0:00
You're watching TBNN.
Search this video
0:03
>> Rain Maker stands accused of of having a role in the Texas floods.
0:06
This is a very very sad story.
0:08
It's on the cover of the Wall Street Journal, not the rain maker part.
0:11
Um that has been contained on X, but I'll give you a little update on what's going on in Texas.
0:17
So, uh Texas Texas rescue grows urgent as toll mounts.
0:21
At least 70 were killed in weekend floods as more bad weather complicates the search.
0:25
Uh the search for swept away for those swept away by punishing flash floods in central Texas over the holiday took on new urgency Sunday as the death toll climbed to 70 and nearly a dozen girls from a private summer camp remained missing.
0:37
Rescuers combing the swollen banks of the Guadalupe River were holding out hope that survivors might still be found.
0:44
The potential for more bad weather Sunday also loomed over ground and air operations.
0:50
The National Weather Service warned of more rainfall and slowmoving thunderstorms that could create flash floods and in the already saturated in the already saturated areas in the Texas Hill Country.
1:01
So uh this blew up on >> and uh and people were asking Augustus did Rain Maker was Rain Maker operating in the area around that time?
1:11
Uh cloud seeding startup RainMaker is under fire after deadly July 4th floods in Texas.
1:18
CEO uh Augustus Jerico who's been on the show multiple times will join us today at noon to uh break it down.
1:23
He's already uh explained his side of the story on Acts several times, but we will ask him a lot more questions.
1:30
He says the natural disaster in the Texas Texan Hill Country is a tragedy.
1:34
My prayers are with Texas.
1:35
Rain Maker did not operate in the affected areas on the third or fourth or contribute to the floods that occurred over the region.
1:42
Rain Maker will always be fully transparent and he and he gives a timeline of the events.
1:46
He says, "Overn overnight from the 3rd and fourth, moisture surged into Hill Country from the Pacific as remnants of the tropical storm Barry moved across the region. At 1:00 a. m.
1:55
on July 4th, National Weather Service, which we work closely with to maintain awareness of severe weather systems, issued a flash flood warning for San Angelo, Texas.
2:04
Note, summer convective cloud seeding operations in Texas do not occur during overnight hours. At 4 a. m.
2:12
on July 4th, NWS issued a life-threatening emergency warning and flooding insured.
2:16
He says, "Did Rainmaker conduct any operations that could have impacted the floods?" He says, "No.
2:22
The last seating mission prior to the July 4th event was during the early afternoon of July 2nd when a brief cloud seeding mission was flown over the eastern portions of South Central Texas and two clouds were seated.
2:34
These clouds persisted for about 2 hours after seeding before dissipating between 3 p. m. and 4 p. m. CDT.
2:38
Natural clouds typically have lifespans of 30 minutes to a few hours at most.
2:43
Even with the most persistent storm systems rarely maintaining the same cloud structure for more than 12 to 18 hours, the clouds that were seated on July 2nd disapping operations in the >> immediately before a massive storm is coming through.
3:06
I think that's the question that a lot of people have.
3:07
Um but we will get into that when he joins the show. >> Yeah.
3:14
I mean there's a big question about how effective is cloud seating.
3:18
Could you start a flash flood if you tried? Um does this work?
3:19
Someone was paying for this because it's not a nonprofit like um obviously state level >> state level.
3:28
So the state might buy cloud seating operations in one way.
3:31
Um there could be you know a mistake.
3:33
He says that he's not involved at all.
3:35
So, we will dig into that with him.
3:39
>> Well, our next guest is here, Augustus Dico, the CEO, founder of Rain Maker.
3:43
Welcome to the stream, Augustus. How are you doing?
3:46
>> Uh, John Jordy, thanks for having me. I am doing well.
3:48
Um, I uh am obviously talking to a lot of people about the flooding that's gone on in Texas and appreciate the opportunity to um clarify that rain maker and cloud seeding had nothing to do with the flooding that unfolded.
4:01
Um and uh even in spite of that uh I think that it's a tragedy that it did happen and certainly don't want anybody to use this opportunity um use this uh controversy to blame cloud seating for the sake of popular political support.
4:16
And you may have seen that Marjorie Taylor Green uh is proposing running a bill to ban all forms of weather modification based on those that we saw in the Florida state house legislature um earlier this year.
4:29
Uh I think it would be both disrespectful to the families involved and baseless uh and without any technical or scientific credibility if that legislation were to go through.
4:37
So um I'm happy to talk about the course of events, what clouding is, what it's not uh here with you today.
4:44
>> Yeah, let's kick it off with um the the high level on what actually happened in Texas, where things stand now, the status of the rescue operations, and kind of the the timeline um that's more broad. >> Yeah, absolutely.
4:58
So um this phenomena, this flooding was global in scope.
5:02
Um it was referred to as a low probability, high impact event.
5:06
Um I encourage people to go to Matthew Kapuchi uh on X. He gave a great outline.
5:13
He's a meteorologist that has a lot of expertise on severe weather forecasting.
5:18
Um but but tropical storm Barry, the remnants of which blew into Texas, was going to cause inordinate flooding regardless.
5:25
And uh that area of Texas is also known as flash flood alley because these events do happen.
5:31
Now 4 trillion gallons of precipitation occurring over the course of just a couple days is pretty out of distribution.
5:37
Um but we are seeing an increase in these sorts of severe climatic events uh over time and especially down and around the Gulf.
5:44
So, just to go over the timeline after having clarified that it was the remnants of tropical storm Barry and the convergence of uh large meoscale phenomena that induced that flooding, um it was at about 1:00 a. m.
5:56
on the 4th that the National Weather Service um issued a flash flood warning.
6:01
Uh and then it was at about 4:00 a. m.
6:04
on the 4th where they said that there was a life-threatening emergency underway.
6:08
life-threatening emergency underway. Um it was not uh I it was over two days prior that Rain Maker had suspended all of its cloud seating operations in Texas because one our forecasters and our meteorologists saw that there was going to be this severe weather event and we needn't operate to produce more water um when there was already the event coming
6:29
but two we suspended operations in accordance with the Texas Department of Licensing and Regulations uh suspension criteria where if there is a uh severe weather warning from the National Weather Service um or there is too much saturation of the soil, we have to ground operations and so we do so both voluntarily and in accordance with existing statutes. >> Okay. So, uh the cloud seeding operation >> Okay.
6:49
So, uh the cloud seeding operation that happened prior to the storm. Uh who was the client?
6:55
Like I mean who I assume someone was paying you.
6:59
Sometimes it's the government.
7:00
Sometimes it's an indivi individual or farmer or business.
7:02
um walk me through uh where they were, who they are, what their goal is by procuring your services. >> Sure.
7:12
So, it it's obvious that at this moment in time, um that region of Texas does not need more water.
7:16
However, throughout the western United States, farms, uh, conservationists, governments concerned with their aquafer supply of water and also reservoirs for both industrial and residential drinking water, uh, contract with rain maker to produce more water via cloud seating.
7:34
And in the case of Texas, the South Texas Weather Modification Association, the West Texas Weather Modification Association, and multiple other uh entities exist as conglomerations of both counties and individual farms that pay for cloud seating services to one,
7:50
water their crops, two fill up the reservoirs that they irrigate their crops with, and three recharge the aquafers like the Ogalala that has been severely drawn down and then puts all of these farmers at risk of not being able to grow, not being able to do business because of a historic trap. >> Okay. So, >> Okay.
8:04
So, what um would the proposed ban just because what what I'm getting at is like I'm wondering if uh like if the government is paying for cloud seating operations like the easier lever might just be to decrease the funding to the government.
8:24
But it seems like Marjorie Taylor Green is pushing for some other legislation that wouldn't just be, hey, buy less of this service because we don't need it and instead this service should never be bought at all.
8:36
So why is there the distinction there?
8:39
Like is is is most of the money that's going into one of these associations uh private farmer capital or is it a split?
8:47
Like h how does that actually break down?
8:50
So right now it's largely public municipal money that is going into these weather modification programs to increase water supply when there is drought or in preparation for drought.
9:00
Um >> the bill that has been forecasted that has been proposed by Marjorie Taylor Green uh would wholesale ban all forms of weather modification be it cloud seating, solar radiation management or what they suppose to be chemtrails.
9:14
I mean, very transparently, I think that a lot of the concern around weather modification is actually conflating baseless notions of chemtrails with a very practical American technology that can and will and does benefit our farmers, our ecosystems, our industrial water needs, and our residential water needs.
9:35
If this legislation were to go through, not only would it deprive all of those interests and all of those Americans from having water from cloud seating, but it would also be against America's interest at a geopolitical level because China recently, I think on the last time I was on TVPN, I talked about how they had a $300 million annual budget for their weather modification program.
9:56
That as of 2025, has been up to$1. 4 billion.
9:58
Um, that is extremely consequential.
10:02
And I think that if we were to ban who controls uh or banning Americans from uh controlling weather modification technology uh that would put us at a meaningful disadvantage.
10:11
Now all of this to say um people deserve transparency.
10:16
They deserve clear regulatory framework so that they know whether modification operations are safe and being conducted in a responsible manner and with government oversight and accountability if ever there are uh negative consequences to cloud seating.
10:32
Again, there haven't been any in the case of Texas, but I think that the reasonable next steps are to more stringently regulate who is allowed to cloud seed, define what the concepts of operation are that are permissible,
10:44
define the suspension criteria at a federal level rather than leaving it purely to the states so that anybody that wants to know about weather modification can look at the data and scrutinize it and ensure that it's being conducted safely. And also just to build
10:56
And also just to build trust because the weather modification act from 1972 that currently outlines uh the weather modification reporting act of 1972 that outlines how we have to report to the federal government is you know 50 years old.
11:10
Um we need more scrutiny on these programs for the sake of public trust and accountability.
11:14
Um and that seems like a reasonable next step that was also recommended by the government accountability office in their report on cloud seating and weather modification earlier this year. Mhm.
11:25
>> What was the scale of the general water mod uh sort of sorry weather modification activities on July 2nd?
11:31
It was you guys a bunch was there a bunch of other players operating?
11:37
Is there generally a lot of players or is it a pretty is it is it a fairly small number of of um kind of service providers uh that are that are participating in these programs?
11:49
Yeah, Jordy, you may have seen uh the prolific hustle on X.
11:52
com posting about this a little while ago.
11:57
He said that I was the CEO of the largest and most powerful weather modification company in the world.
12:02
Um and >> I saw somebody compare somebody was comparing weather modification tech to being saying it was more dangerous than nuclear >> nuclear bombs. That was kind of crazy. Yeah.
12:14
>> And then I also saw some people just showing like general flight logs of like commercial airplanes.
12:18
Like obviously there's a lot of >> people have every right to be angry and demand answers.
12:24
It's such a tragic >> Yeah.
12:28
>> incident but but yeah I'm curious to get into the the scale of of you know kind of maybe late June early July what was going on broadly. >> Yeah absolutely.
12:38
So there's one other cloud seating operator in Texas called uh seating operations and atmospheric research soar.
12:45
They're responsible for operations over the rolling plains uh weather modification association which is significantly farther northwest of Kirk County.
12:54
Um on July 2nd we conducted one 19minute cloud seating flight where we released about 70 g of silver iodide and 500 g of salt table salt.
13:04
um that was released at about 1,600 ft above ground level into two clouds that dissipated over the course of 2 hours after seating them.
13:14
The amount of time that those aerosols could have been suspended in the atmosphere is less than the time between when uh we were seeding and the onset of rains from uh the remnants of tropical storm Gary.
13:27
and the amount of material that we dispersed could not come anywhere close to inducing the precipitation, the 4 trillion gallons of precipitation that did come from that event. >> So, yeah.
13:40
>> And and I'm assuming you guys like have records or you keep records of like the radar showing these different cloud formations.
13:49
So you you're you're it's not just we looked and we think it dissipated, but it's like you can actually you have like you know basically a a map that's live updating.
14:00
Is is that the right way to think about it?
14:02
>> Not only do we keep records for our own research purposes and operational purposes, but we're required to keep records by the Texas Department of Licensing and Regulation.
14:13
And those are accessible online as are the reports on our seating activities.
14:17
And if anybody is interested in those, then you can ask for them from the TDLR. >> Um I I'm I'm curious.
14:22
Um when when the the flooding happened in Dubai, I want to say it was a year or two ago.
14:27
Um Dubai is known for their cloud seating operations. It's very dry place.
14:33
Uh and makes sense why they would want to uh increase precipitation.
14:39
A lot of people uh maybe the same types of accounts that have been that have been blaming you were quick to blame it on cloud seating throughout history.
14:50
Has there ever been any major kind of flooding event that that people were able to say yes 100% this was caused by weather modification activities >> or is the tech not even powerful enough yet to to do something like that?
15:05
So I I think that there's probably three points to touch on.
15:11
Um the first of which is that it wasn't until 2017 that attribution had been uh physical attribution of cloud seedings effects had been seen and proven in an academic context.
15:24
And so with new advents in radar technology, namely dual polarization radar, we're able to much more clearly monitor what the effect from cloud seating is.
15:33
In previous operations, it was extraordinarily difficult to see what your effect was because we could not measure the cloud dynamics uh and the cloud microfysics that were changing as you were seating.
15:46
Um so that's the first point.
15:46
The second point is that and again I'm trying to be and will continue to try to be maximally transparent about our operations and historic weather modification.
15:55
Um there was something called Operation Popey during the Vietnam War where the deliberate intention of cloud seeding was to cause precipitation that would uh like cause flooding and then impede supply chains on the Ho Chi Min trail.
16:12
Um >> now the extent to which that was effective because we didn't have good satellite imagery or dual pole radar is outstanding.
16:18
Now that said, lastly, third point, we have suspension criteria that are given to us not just by the TDLR in Texas, but every state wherein we operate because if there already is too much saturation of the soil or if there is uh an oncoming severe weather event that the National Weather Service has uh notified us not to seed, then we ought not do that to increase the severity of precipitation.
16:44
severity of precipitation. So there are suspension criteria because there are limits on what we ought to do with this technology um so as not to cause flooding and only reap the rewards from it right for our farms for our ecosystems and for our national security interest as well right like if we don't have access to weather modification
17:02
technology if we don't regulate this at a federal level and ensure that there's accountability and attribution for these activities then other people other nation states could be conducting weather mod in the vicinity of or on American soil without any accountability and so that's why I am advocating for way more regulatory scrutiny from the federal government for cloud seating and weather mod ops. >> Uh walk through some of the history of
17:24
>> Uh walk through some of the history of the the Chinese weather modification uh strategies.
17:29
Um we we heard about the the the flooding in Dubai that was kind of unclear.
17:34
Have there been any notable or confirmed negative outcomes from China spending I mean you said $300 million a year, something like that.
17:44
that that seems like a lot of cloud seating.
17:46
Seems like if there was a surface area where there could be mistakes made, they would have kind of explored that.
17:51
Um I remember the the pre-Olympics they were doing cloud seating or just kind of bringing down like the the dirt in the atmosphere.
17:59
Um and you know people kind of learn from that.
18:04
Okay, you get acid rain when you do that uh in in in particular.
18:06
But uh have there been any case studies from China that uh we should be learning from in America?
18:12
um case studies from China with adverse weather coming from their cloud seating operations. >> Yeah.
18:20
Anything like that like like something where like okay they they've done a lot of this push this to the limit.
18:25
They've put they've done this at scale.
18:26
If there's going to be rough edges or mishaps, I would have I I suspect that we would have seen evidence of that over there.
18:34
They would have had an accidental flood or something like that happen over there.
18:38
If they're doing it at scale, >> you would expect to have seen it from China.
18:44
Um, however, you would also probably expect and understand that they're a relatively inscrutable country that does not report on their activities very uh openly and objectively.
18:53
Um, now that said, one thing that we do know about the weather mod program that they do have going is that they're planning to buildund 100,000 ground generators on the Tibetan plateau. Mhm.
19:07
>> So, Rain Maker uh is primarily using drones for our operations.
19:10
Uh we also have inherited some ground generators from previous operations.
19:14
These are essentially um aerosolizing units on the tops of mountains.
19:18
They can disperse material into clouds uh when the clouds intersect those mountain tops themselves.
19:24
>> Is that like a cannon that fires the material into the cloud or >> No, no.
19:27
You you might recall my my initial inclination to use something like that cuz it is used in China.
19:31
Um, but no, it's it's essentially like a uh >> uh a smoke stack of sorts, a very small smoke stack that releases those aerosols there.
19:40
But in building uh a 100,000 of these ground generators and also using the Wingong 2 and a bunch of their other military drones for aerial cloud seating, um they're turning Tibet into uh a reservoir, a a snowpack reservoir of unprecedented scale that will feed more water into the agricultural basins in southern and eastern China.
20:00
And I think that uh you know although again this is something that needs to be transparently reported on and regulated um depriving American farmers in the west especially as a congressperson from Georgia right where there is not as severe uh reliance on cloud seating to produce water would be against America's interest. Mhm.
20:24
>> Jordy, >> I guess >> I'm I'm trying to I mean the the the >> my question uh is it feels like it it feels like candidly it will be hard to come it'll be hard to find uh any type of allies uh in Texas on the ground in Texas maybe aside from from the farmers but but I'm curious um you know the the the various different groups you know what
21:01
what the reaction from them has been >> in terms of you know if they're you know it's the the reality is is um water scarcity affects all every person in Texas but only a few people truly feel it right it's a much smaller group because everybody goes to their sink they turn on the water they turn on a hose outside they go to a grocery store, there's water, there's produce. It It's
21:24
It It's not something that people necessarily feel.
21:29
And so I'm curious where um you know, you obviously are going to defend weather modification um because you you believe in in the many different ways it can have a positive impact, but I'm curious uh who you think uh the other players that will will be on your side as the industry I mean the industry was not in a good spot prior to this.
21:50
it's in a much worse spot um now and I know you've been flying all over the country making sure that it doesn't get banned.
22:00
So um I'm curious what what you think the kind of coalition that will kind of form uh around you. >> Yeah. Yeah.
22:07
Well, so I I actually think I uh just from my own experience over the course of the last few days disagree with the two points that you made, right?
22:16
like it it has neither been hard to find allies for cloud seating weather modification in Texas nor do I think the technology and the industry is positioned worse now than it was prior to this weekend.
22:28
Um, and regarding the first point, uh, there are some people that I think, um, are probably not in good faith engaging with this because they have some preconceived notions about chemtrails or otherwise, um, and don't themselves want to scrutinize the data to back up how our operations are different and beneficial, uh, whereas chemtrails, as they believe them to be, are, you know, uh, malevolent.
22:52
Um, the vast majority of people that I've interacted with online, on the phone, and in person are rightfully curious, skeptical, concerned.
23:05
some you know uh more than others obviously but in scrutinizing the data and having these conversations and learning about what cloud seating is pretty unilaterally people are supportive of it provided that there is a regulatory framework more stringent than the one we have now that ensures that it's safe.
23:24
Um this is true both of just individuals um that are not themselves farmers but obviously farmers, water managers, uh government officials too.
23:34
Um I welcome any questions that people do have both online and via email about what our activities are, what our policy recommendations are.
23:44
Um, and and I'm I'm grateful that there are a lot of people that understand one, our operations did not contribute to the flooding, but two that even if there was a flood now, it doesn't mean that there is always enough water and having access to a technology to produce more water for farms and otherwise uh would be beneficial.
24:01
Like people want a more green, lush uh country. Um >> yeah, I'm curious.
24:08
Uh I'm sure you've spent plenty of time thinking about this, but is would there be a way to apply the existing technology you have almost in a defensive way >> in you know theoretically uh >> see a hurricane while it's still offshore >> something like that.
24:26
uh or the or or you know one of the issues here there was just so much water in the atmosphere that rolled over a heavily you know populated area and then it's got it's it's gravity right it's got to come down >> um you know is there an application of the technology that could over time strategically prevent you know or or act defensively against the conditions that create flash floods.
24:52
It's it's a very worthwhile question for you to ask and for us to ask ourselves collectively.
24:58
Um right now again rain maker only does precipitation enhancement operations for all those constituencies that I listed before.
25:08
However, in the past the United States government funded uh project storm fury which was a series of attempts to reduce the severity of hurricanes over the Atlantic before they broke against the eastern seabboard.
25:19
Um again, we didn't have the appropriate understanding of atmospheric science or the radar or the satellite data necessary to appropriately do that.
25:28
However, uh severe weather is something that is like a geopolitical risk, a national security risk.
25:34
Um it causes damage and it is fundamentally a physics problem, right?
25:39
A physics and chemistry problem.
25:41
Is there technology now that could mitigate severe weather like this? Um no.
25:45
And Rain Maker doesn't have it.
25:47
Is it possible to someday provided we invest in Noah in the National Weather Service in the appropriate research into cloud seating such that we could reduce the severity of severe weather? Absolutely.
25:59
And I am entirely in favor of that provided it is done in a responsible manner.
26:04
Um, and if we were to ban it wholesale, then not only would we lose access to precipitation enhancement, but we'd lose out on any potential of at the very least better forecasting for these systems and warning people early, but also the even greater and more consequential beneficial potential of reducing severe weather in the future.
26:26
And so I think that the United States government and Rain Maker should and and are absolutely interested in mitigating severe weather in a manner similar to Project Storm Fury. M that makes sense.
26:37
>> I I I think the PR what you were getting at Jordy like the PR difficulty here is that like when there's not enough water, >> crop yields are lower, prices go up, but it's very distributed.
26:49
Everyone feels it a little bit.
26:52
Whereas when there's too much water and there's a flash flood and individuals die, you have a very it's a very emotional, very uh it's very concentrated.
26:59
The pain is very concentrated.
27:01
And so that's why this this story Normally when there's a natural disaster, >> there's >> you can you can critique the government for their response to it, >> but there's not somebody sitting there that a scapegoat, right?
27:16
And so I guess the question is it's easy.
27:18
Yeah, >> it's it's you know whe whether it's online accounts that are just engagement farming >> or it's a politician >> uh you know >> scape you know the the concern is that uh and your concern is that the industry becomes a scapegoat and uh America loses a capability that our adversaries clearly care a lot about. >> Yeah.
27:42
My my question is like we're we're seeing this bifurcation.
27:45
It seems like Ted Cruz came out in support of the idea that cloud seating had nothing to do with the Texas floods.
27:51
Marjorie Taylor Green is taking kind of the other side of that.
27:54
Um my question is like these are politicians at the end of the day.
27:59
They're not independent scientists. Who can we go to?
28:02
Who can the population go to for like a truly independent review of this situation?
28:09
like is there is there some sort of independent governing body or are there are there respected scientists that kind of don't have a financial or you know political incentive one way or another?
28:22
Um h how do you think the uh the the populace should be satis obviously you're telling your side of the story you're going direct you're explaining things you're laying out the data but what uh what what do you expect people to look for in an independent analyst? >> Yeah. Yeah.
28:41
So, for one, um I think that Noah, the National Weather Service, the National Center for Atmospheric Research, um >> all of those are great third party entities that can review the information, corroborate the information that we've provided.
28:59
>> Um pro >> provided of course that they continue to exist and remain funded.
29:02
Um, I think that this probably demonstrates why it is important that we should retain some capability nationally to forecast and research the atmosphere because there's there should be some body that's capable of reviewing this to ensure that it's safe.
29:18
>> Um, I'll also say, you know, regarding the uh scapegoat dynamics that that exist right now, um, I've thought about this >> pretty prayerfully and intently over the last few days.
29:30
And when there is a calamity of some sort, like I I've been trying to think about why people are uh say coming after Rain Maker or uh angry at Rain Maker.
29:41
And I think that when there is a calamity of this type, um if there was someone responsible, if there was someone or something that could be held to account, then in holding them to account, uh you could supposedly prevent this kind of thing from happening in the future.
29:57
Um the trouble with a true natural disaster as this was is that there is nobody to be held accountable.
30:05
Um and that makes the world a lot more tragic because it means that things like this will persist.
30:10
Um they they will persist indefinitely into the future.
30:13
Um unless and until some sort of technology could reduce the severity of severe weather.
30:22
>> Um we went through this with the California fires.
30:24
You know, it was like everyone was searching for like a single person to pin it on and like it came down to like, you know, some people built their houses the wrong way and there's some building codes that need to change and there's some water rights and water flow and there's some different like we need more goats in certain areas.
30:43
There's like a million different things that could have prevented this if they all were all working together as a welloiled machine and had the forethought.
30:50
Um, but it's a very very frustrating and difficult situation.
30:52
So, our our thoughts and prayers are with everyone who's been affected.
30:56
Um, but thank you so much for stopping by. This is fantastic.
30:59
Thanks for uh breaking it all down for us. >> Thanks, guys. Appreciate you. >> Cheers.
31:04
>> We have some maybe terrible news.
31:04
There might be top signals in the market.
31:10
There might be top signals all over the place.
31:12
>> We've been uh building out an internal top signal tracker, crowd sourcing some of them. >> Crazy. >> And it's a long list.
31:17
We'll get through At the top of the list, podcasters have been wearing white suits recently to celebrate the market ripping.
31:24
That feels like >> white suits are actually a top signal.
31:28
>> It's a complete top signal.
31:28
Um, but of course there is some good there are some the economy is strong.
31:34
We're going to go through uh Joe Weisenthal's breakdown.
31:38
Things are not doom and gloom, but there's a lot of crazy stuff happening and it's fun to dig through.
31:42
Uh, I mean, the first major top signal, Bitcoin, alltime high. >> Yep.
31:47
>> Yep. you know that's always you know it is definitionally a top signal because >> let's go through the list here because some of this is quite substantial >> some of this is kind of anonymously contributed through uh group chats uh
32:00
some stuff we've we've observed uh we're going to catalog it and see if we can turn the tide of the top signals to okay ideally >> so starting off uh yesterday >> uh Trump made a post on true social calling uh basically celebrating the
32:16
state of the economy the markets uh you know really calling out how how many assets are performing well the second one you want to read through it you want to read through it a little bit because he basically >> get through the post and and we'll get to the moment
32:30
>> so Donald Trump on truth social truths uh tech stocks industrial stocks and NASDAQ hit all-time high record highs crypto through the roof Nvidia is up 47% since Trump tariffs USA is t taking in hundreds of billions of dollars in tariffs. Country is now back. A great Country is now back. A great credit.
32:51
Fed should rapidly lower rate to reflect this strength.
32:53
USA should be at the top of the list.
32:56
So low rates are low rates are actually just a reward for when the markets are ripping.
33:01
It's a little treat that we give ourselves. >> Exactly.
33:05
>> When things are great. >> Yep.
33:06
And the White House is posting this screenshotted on Axe.
33:10
The country is now back, says President Donald Trump.
33:13
every account controlled by the White House has been on a tear.
33:18
Some of them uh some of the posts I think are a little bit um low class and vulgar, but others others are quite funny.
33:27
>> But there's definitely the memeers are in control.
33:29
Didn't Ruin say every like uh every politically aligned um poster he knows who is like pro Trump now works for the White House, but like you just haven't seen it cuz they were like a nons and they just kind of dropped off posting and now >> they be getting death threats.
33:44
So they have to it's even actually more in many ways it's it's more controversial work than Doge. >> Maybe. Yeah. Yeah.
33:50
Maybe it's more under discussed because Doge had this big like question in the media about like, you know, is Elon doing something that's, you know, he shouldn't be?
33:58
Is he a government employee?
34:00
Like what's the relationship between the two?
34:01
Uh, and so, you know, there was a lot of investigative journalism that went into figuring out what's going on with Doge, who's involved. >> Yeah.
34:09
Nobody's investigating the memes.
34:11
>> The social media managers, which is >> they need to be investigating the me the memes of production.
34:14
But, um, >> anyway, so so going through uh my list here. That's great.
34:18
Uh Eric uh Trump a while back said this is a good time to buy.
34:23
This was a few months ago on on Ethereum >> on Ethereum.
34:27
>> And then it just went down for months and now it's back up and now it's back up and he's saying you're welcome.
34:32
>> I do remember Trump he called the bottom, right?
34:34
He said like now is a good time to buy generally and then the market ripped since >> he created it and he and he called it perfectly. >> It's wild. >> It's finesse.
34:41
Um more going down the list.
34:44
uh Coinbase just uh who who we love, but they did uh they're a Fortune 500 company.
34:49
They did update their profile picture to an NFT.
34:52
Historically, that has been a top signal.
34:55
I do think their profile pictures you have any experience with NFT profile pictures.
35:01
>> You know, I've delved I've delved over the years. >> Yes.
35:05
And if you and if you look at maybe the moment that I did use an NFT profile picture in 2021, >> uh it was maybe only off by one or two months in terms of in terms of the >> I never used a NFT profile picture, but I bought an NFT >> right near the top. >> A chain runner. >> Chain nice.
35:22
>> Which I still own, >> which actually I didn't like overinvest, get over my skis as very small portion of >> it's an asset that will be passed down through your family like a fine watch.
35:29
I like to think of it as like a piece of 2022 lore, you know, it's just like a piece of history.
35:36
Uh but uh yeah, yeah, fun project.
35:40
And I feel like to some degree, you know, you're not really It's like a skin in the game question, like you're not really participating, you're not experiencing the the market unless you're participating to some degree, but you don't want to get over your skis.
35:52
Um also did the NFT profile picture at a really bad time and had to roll that back.
35:58
Like there's been a number of like NFT profile pictures that have been like >> it is a historical top signal.
36:01
>> it is a historical top signal. It could be it could be now just a signal for the start of a you know generational run new cycle but uh historically it's a top signal so we got to call it out >> if if NFTTS are going to make a comeback
36:15
because like there's been like crypto has been coming back and Bitcoin went from what 30 >> will be back when A-list celebrities are using them on their Facebook accounts >> that was a >> that's the real test X account >> could see happening early. >> Yeah. >> Yeah. >> Facebook account.
36:36
>> Original Facebook account.
36:37
>> There's got to be a new project then cuz I don't think any of the old pro old old products or projects are going to, you know, come back. That would be crazy.
36:46
Although some of them are kind of lindy like haven't the original >> Crypto Punks.
36:49
>> Crypto Punks, those have kind of held their value.
36:50
But the board apes have sold off like crazy, but are still it's it's unfortunate.
36:54
board apes are not in gag gift territory yet cuz you you think oh it' be funny to get like your buddy like a board ape for their birthday >> 30k or something >> but it's like yeah it's like >> Tyler how what's the floor price of uh of board apes I'm I'm interested to know um while Tyler looks that up let me tell you ramp time is money save both easy to use corporate cards bill payments accounting and a whole lot more all in one place go to ramp.
37:16
com also we don't we never shout this out 4.
37:17
8 eight stars on G2 with over 2,000 reviews. That's great.
37:22
Shout out >> world class. Okay.
37:23
Another >> the floor price is like around 10 ETH.
37:27
So that's like almost $3,000. >> 3,000. 30,000. >> 30,000. Yeah.
37:31
That's like not a gag gift.
37:36
Maybe for the man who has everything. >> Yes.
37:38
For the man who has everything. Great uh great gag gift.
37:42
>> It is pink elephant at Sun Valley.
37:44
But that you know by by you know Christmas time >> do they do pink elephants at Sun Valley?
37:50
I feel like they should >> maybe maybe ask some of our some of our friends that are there this week.
37:55
Uh so in other news um >> uh Robin Hood CEO Vlad is raising uh at $900 million valuation for a math foundation model startup.
38:06
And Vlad and Robin Hood have been on a pretty generational run.
38:11
But this does feel uh a bit uh top signaly right uh especially in the context of grock oneshotting PhD level math uh in the announcement on Wednesday.
38:26
So interested to follow that one.
38:28
Uh optimistic but again >> mathematical super intelligence.
38:32
>> Historically when we've seen CEOs of of public companies start ripping you know second companies and then and then uh getting these types of valuations uh without a lot of underlying revenue uh it can end uh poorly.
38:48
U >> uh Andrew Wilkinson is giving uh stock tips.
38:54
He hit the timeline today. I'll read through it.
38:56
He was highlighting a company uh historically a value investor.
39:02
>> Uh but this morning things like the Warren Buffett stuff, right?
39:06
>> The um Birkshshire Hathway for the internet. >> Yeah, that's right.
39:09
>> He says, "There are many ways to profit from the AI boom, but my favorite is iron. I rarely buy stocks.
39:13
The private market is way too attractive, but every once in a while, I see something something that stops me cold. In 2025, it's Irend.
39:21
I call it a Picasso I found at a garage sale.
39:26
>> The stock is up 54% since he recommended it on my first million, but it's still cheap.
39:31
Here's the trade in a nutshell.
39:34
US capacity for energy and compute is highly constrained.
39:36
Two, permitting and building facilities takes years.
39:38
Three, AI scaling laws are continuing to deliver, but even if they don't, tons of compute is required for inference. >> Mhm.
39:47
Iron is a highly reputable, publicly traded Bitcoin miner with massive data centers mid build in Texas.
39:56
It pivoted away from mining Bitcoin at these new facilities to instead build them out for AI training and inference.
40:02
Once completed, these facilities should generate in the range of $2 billion in new cash flow.
40:07
>> This company's name IN.
40:07
Even if AI completely fizzles, these facilities are highly valuable as traditional data centers or can be rolled back to mine Bitcoin.
40:15
So, it's an AI thesis, but if AI doesn't work out, we can still mine Bitcoin.
40:20
The entire market cap is currently 3. 8 billion.
40:22
So, uh Andrew, I don't think this is investment advice, but it sounds like it.
40:30
Um and uh interested to see uh see where this one goes.
40:37
But uh anyway, anytime you see a value investor start trying to cash in on the AI boom should be a little bit wary.
40:46
>> Uh Harry Stabbings today was calling >> not like like it doesn't have earnings, right?
40:50
It's a lot no it's it's trading around $4 billion.
40:54
>> I don't think it's ever generated any profit.
40:57
>> I mean it says it says 23 million in EBITD DA but in 2024.
41:02
So I I don't think it's like losing that much money.
41:05
And I guess net income in the last quarter was 24 million, but the net income to market cap ratio there is 40 I guess. So still pretty high. >> Yeah.
41:16
I mean the the the thing here is at the same time SATA is pulling back on new data center development.
41:22
He's happy to be a leaser.
41:25
>> You have incredible neo clouds that have deep domain expertise.
41:28
deep domain expertise. the Iron team I don't think has a bunch of team around uh running large AI training or inferencing >> and so um anyways >> just feels like they're a little bit late to that party because there's already like three or four did I make the uh cluster max uh Dylan Patel article >> I doubt it because they're not online yet right >> oh sure sure >> yes because semi analysis does the uh
41:56
cluster max rating for all the neoclouds including the the hyperscaler clouds And I feel like they did not have um let me see iron I don't think is on here tensor wave there are so many runpod lambda scaleway smc azure nebus together cruso
42:13
leptton oracle coreweave AWS >> so hyper competitive market unclear if this Bitcoin miner is going to be able to pivot into AI training and inference uh in this >> when they're up against the uh players that you just mentioned. Uh, another top Uh, another top signal.
42:31
Um, I'm not going to go out and say that this is impossible, but Harry Stebings is calling for 8 trillion Nvidia in the next 5 years.
42:39
Uh, private markets investor backed a bunch of unicorns.
42:44
Um, uh, starting to make, you know, very specific uh, sort of price predictions uh, on the timeline.
42:54
>> The specificity of the price prediction is interesting.
42:55
I was thinking about that like should like as we talk about tech companies should we be trying to like boil down to like price targets and I just feel like that's not the domain of of talking heads necessarily or like podcasters I guess >> or private markets investors. >> Yeah. I Yeah.
43:13
It's just it's just hard because like to do a proper price analysis on a big public stock like you you really have to look at the financials like you have to read the the the the financial reports.
43:22
you need to actually understand the underlying financials.
43:25
Um it it like the vibes based analysis doesn't seem appropriate usually, but um >> who knows?
43:32
I mean sometimes vibes are all you need, John.
43:34
>> Yeah, it's certainly been like it's I mean when was when was Nvidia $2 trillion stock?
43:38
Like when was last doubling?
43:40
Like in the last year or something? I don't know.
43:41
We can pull up the video chart. >> Well, uh moving on.
43:44
We another incredible top signal.
43:47
Circle, a great American uh stable coin company, is trading at a 2,300 >> uh PE ratio.
43:57
Uh nearly uh at once, I think they eclipsed Coinbase's valuation very briefly.
44:06
>> Despite the fact that they give half of their revenue to Coinbase Y >> as part of their distribution partnership.
44:12
So again, lots of excitement around stable coins.
44:14
um feels like Circle could potentially be uh a little over its skis, but it's a great company and they have a lot of advant advantages now.
44:22
But the um it's very euphoric uh uh multiple.
44:25
Another top signal we have is Sohan Periq.
44:29
We had him on the show >> uh just a week ago.
44:33
Uh this same sort of thing was happening in 2021 2022 where engineers were really ramping up moonlighting activity, right?
44:41
they be working at meta and then working at some startup or things like that. co maybe accelerated it.
44:48
But again, if if companies are so desperate to hire great engineers that they'll run these like super fast hiring cycles, put up with people generally talented people that are underperforming, right, which Soh was was not delivering, was making a lot of excuses and a lot of people rightly let him go quickly. >> Yeah.
45:08
It's just a it's just a the nature of like the dynamic of uh just competition.
45:14
Like if your competitors are hiring really fast and you need to hire really fast, you're just like, "Okay, well, we don't need to go deeper.
45:21
So with let's wind up fasttracking this person."
45:23
So you wind up hiring uh you know, the same person five times. I guess >> it happens. It happens. >> It is.
45:29
It is just like a funny anecdote that like is like oh wow those were some pretty crazy times. Remember that anecdote? Remember this anecdote?
45:35
It feels like we're we're in this. >> Moving on.
45:40
Uh MASA top blasting or potentially top blasting.
45:43
Uh anytime MASA historically MASA getting into the headlines, uh whether that's Stargate, >> Yeah.
45:51
>> uh structuring this $30 billion investment where nobody knows or and the 500 billion, nobody really knows where the money's coming from.
45:58
>> They're exciting big headline numbers, but unclear if uh he will actually be able to deliver on that.
46:04
I think him trying, you know, getting in the breakout, one of the breakout consumer AI winners, which is OpenAI, is smart.
46:13
He should have exposure there.
46:13
Um, but I think everybody should be a little bit uneasy that he's pulling out the checkbook >> and and writing numbers of that size.
46:22
Um, >> also also investing in not just OpenAI, but like a new company that is a data center holding company that may not have the same economics as OpenAI.
46:29
So, there's a big question there about like how much he deploys.
46:33
I I'm trying to remember the uh I mean we did that whole deep dive on Masa and you know he made a ton of money on AMD but that when he made that investment it was like a way less frothy time or you know it wasn't AMD it was uh it was uh what what what was the Soft Bank uh chip deal? >> ARM.
46:52
>> ARM yeah when did that ARM deal happen?
46:56
Um SoftBank require uh owns roughly 90% of ARM.
47:00
They acquired in 2016 for 32 billion and later took it public in 2023.
47:07
I'm trying to think 2016 was that a particularly frothy time for him to get into that deal because he has he has done a number of really great deals.
47:14
But when >> like the other one is the other one is is Yahoo.
47:20
You remember he he had this crazy meeting with the Yahoo team >> where he basically was like take my money or I'm going to and he was like didn't he ask he was like who are your competitors?
47:30
>> I'm going to give money to He didn't even know who the competitors were, but he said, "If you don't take my money, I'm going to go give the same check to them."
47:36
>> Uh, so he they ended up taking it.
47:36
He acquired approximately 41% of the company uh at um somewhere around a $200 million valuation.
47:48
>> Uh when Yahoo went public in 1996, the uh he had an instant paper profit of 150 million.
47:57
But then at the peak of the.
47:57
com bubble, Yahoo was valued at 125 billion.
48:03
So anyways, uh phenomenal investment but uh very different uh valuation and and ownership targets and and uh unclear.
48:10
Um I would love to see OpenAI get you know for profit and get public uh but uh for to to you know we'll we'll have to see.
48:22
Um, going down the list, another classic Pomp spack that uh we we had Pomp on the show to talk about it. >> Spacks are back. >> Spacks are back. Pomp's got a spack.
48:31
A lot of people were calling that a top signal.
48:35
I I'm excited to see what what Pomp does uh with with his.
48:37
Um but in general, this uh >> this extreme excite retail excitement around these sort of Bitcoin treasury companies is fascinating. >> Yeah.
48:50
in the context of it now being very easy to get Bitcoin exposure in a variety of different ways.
48:56
I I'm not sure we need a bunch of net new Bitcoin treasury companies.
49:02
Yeah, it's it's it's mostly that like whenever there's a whenever there's a new trend or bubble there like there it's very easy to map like okay there's one company that it's really working this is massively successful like everyone is using chat GBT like AI is a thing it is it is real the internet was real Google was real Amazon was real but the the 25th Amazon copycat did not do well.
49:32
And so that's always the risk is that you've applied like the same overarching theme to something that's like so far down the power law that it will never grow into the valuation that it's been assigned. That's always the risk.
49:47
>> Y >> what else do you have?
49:48
>> Darkh updating his timelines. Uh that happened Monday. We had him on the show.
49:54
It was uh it was fun conversation.
49:54
I think Dwart Kash has remained incredibly bullish and and uh and I think he rightfully is.
50:03
He also uh is being somewhat of a realist and being like I don't think that AI is priced in to the market broadly.
50:11
But I do think that some of the promises of AI will take >> another couple years, another 5 years, etc.
50:19
to really deliver versus some of the much more hyperaggressive AI 2027.
50:25
You might say that AI 2027 itself was in in hindsight that could end up being like the number one top signal which is that >> basically if if you haven't read the the >> uh kind of study paper essay um they basically say that by by 2027 you know a
50:42
single foundation model company could just be acquiring every auto manufacturer in the US to develop you know uh millions and millions of robots that would then you build uh you know and and we would hit this sort of fast takeoff. >> Meanwhile, Apple is like we can't
50:58
>> Meanwhile, Apple is like we can't possibly get out a slightly lighter uh VR headset until 2027. >> Yeah.
51:05
>> Like and and this is what we do like like we make stuff.
51:08
>> We've been working on this for a decade.
51:10
>> We make stuff like every year. We are the best at it.
51:12
We make the most stuff and the best stuff pretty much.
51:13
The most complicated stuff.
51:15
That's what we make when we're in the widgets business.
51:17
And yeah, making that headset lighter, it's going to take us a full 2 years to refresh that. >> And I liked 2027.
51:24
It was a It was a fun reading. Very thoughtprovoking.
51:27
But uh I I think that uh we will be uh we'll have to circle back on it 2030 or even 2027.
51:37
I mean the the the the big thing was you know our conversation yesterday with with meter um about the actual like are we are we close to reinforcing AI uh where the AI models are self-improving and and I was kind of you know like okay I I really hadn't read the full report beforehand so I didn't really know what to expect.
52:00
to expect. I was blown away because uh I was expecting you know you know something between like you know like ARGI it feels like with ARGI we're 10% towards solving something there which is just like you know a basic versatility in AI um that it can solve things that humans can solve and it's not narrowly defined it's generalizable now uh RKGI is like the perfect example of like we maybe haven't hit we've done
52:29
intelligence but we haven't done general intelligence yet and everyone keeps saying oh this is AGI that's AGI and ARKGI is really holding it back saying like well if it was truly general should probably be able to solve this basic
52:42
puzzle that a kid can solve um and and for that it's like okay we're going from like 9% to 15% like we are still like you know 85% in not even like you know nowhere close um and uh and the uh the um the the the meter report. I was
53:01
I was expecting it to be like, well, you know, yes, we're seeing, you know, uh slight gains on self-reinforcing AI development and the and the the AI is starting to help build the itself slightly and and the result was like, no, it's actually setting us back in in in this domain. It's not working at all.
53:24
And so that was like a pretty pretty big like okay there's a there's a completely different like not that it's not useful the stuff's useful all over the place.
53:32
I saw Rune talking about that he was like for so many different projects it is useful but for the frontier like it's not the product that's advancing the frontier at all.
53:42
Y >> but yeah I mean that that that probably bridges into the the talent wars but >> well yeah bridging in uh I do think that in hindsight uh we will look back in maybe a year, two years, 5 years, 10 years and think about the signing bonuses and general offers of AI researchers in June and July of 2025 as being somewhat of a uh top signal.
54:05
I think it is very strategic and makes sense from Zach and Meta's point of view, right?
54:13
When you look at their AI capex, >> it makes sense for them to have the best possible team and they have the balance sheet and the general profitability in order to uh do something like that.
54:23
But in general, uh, AI researchers who, you know, 6 years ago, uh, didn't get any attention, uh, much attention at all from the media.
54:35
The fact that they're now trading for more than NBA superstars, more than, um, more than, uh, you know, Tim Cook's annual uh, total comp um, >> it will be an obvious one in hindsight.
54:49
Uh the other one uh $6 half billion dollar aqua hire of of IO.
54:52
Uh I think that again you can rationalize it in the sense that it's a couple points of open AI to put together the best founding >> hardware engineering team probably in the world that's available collectively.
55:09
>> Uh but at the same time again it's it's quite a lot uh considering you know the company was barely I think a year old um at the time.
55:17
Yeah, it's it's interesting because like chat GPT is so it's so installed like it feels like it's already Lindy and it feels like even if there is some massive correction like in in the market or in AI generally or some pullback like people are still going to be using CHP as an app right in the same way that Amazon made it through the crash.
55:39
Uh the question is like what what will it take for the IO acquisition to look like the Instagram acquisition in hindsight?
55:47
Like they still kind of have to go from 0ero to one with that project which is very different than Instagram which was already a mature and growing business.
55:56
It wasn't >> they figured out ads really well.
56:00
>> Well Instagram they were were they doing ads?
56:02
>> They weren't doing ads. >> Oh yeah.
56:04
>> Saying but Meta was like we know how to make >> Yeah.
56:06
It was like a perfectly complicated perfectly complimentary business.
56:09
>> We know how to monetize social users better than anyone on earth >> and you have gotten a bunch of users and it's working and it's growing and and you're even >> and we can actually accelerate the growth of the business in a bunch of different ways.
56:19
So it'd be very different if it was like, okay, yes, IO is selling, you know, like like it's it's a small but growing hardware company that people love >> with a product people love >> with a product people love and maybe they can't manufacture enough of it or maybe they're maybe they're underetizing it right now, but people love it.
56:34
But it's like it's pre-launch >> like Yeah, multi-billion dollar acquisition for pre-launch is pretty crazy. >> Yep.
56:41
Uh going down the list, um what else do we have?
56:45
I think I think the tokenized private company shares I I think it with um without you know Republic and and Robin Hood both creating products that are completely unauthorized basically der derivatives the companies that they're they're offering are are angry at them saying don't do this.
57:04
>> Um and uh >> it's a Spider-Man meme of like top signals pointing at each other.
57:09
Everyone's like this is a top signal.
57:12
>> Anyways, I'm excited about these experiments.
57:13
I just think that um uh I'm a little bit wary.
57:16
Uh and then last but not least, Satya doing two rounds of layoffs this year.
57:23
Uh M we've tal we've reported on this before.
57:25
Microsoft does routine layoffs.
57:27
I think they're pretty good at at kind of identifying underperformers or people that should just move on to different different roles.
57:34
Um but uh Satia I think has been I think will look back and he's been uh excited uh but pragmatic right um and uh I think that uh he will u when the dust settles I think he'll look pretty good.
57:51
Yeah, I wonder like if there's some massive pullback and I mean I I don't even know what what what that would look like essentially like if let let let's assume that the the current capability of AI models essentially plateaus for like a decade or something like that just hypothetically.
58:09
Um and you know they're useful but it's not some reinforcing fast takeoff super intelligence.
58:17
What is Microsoft a big loser in that scenario?
58:20
It seems like Satcha is pretty well positioned, right? >> Totally.
58:23
>> Um like the company prints cash is very healthy, has done these layoffs.
58:27
They'd have to retreat from some stuff and some of the promises that they made maybe.
58:34
Um, but in general it seems like they'd be really really well set up to just like like stick through.
58:37
But I'm trying to I'm trying to think of going back to the the.
58:41
com bubble and and the like you know the effect of like Oracle's mainframe business like probably made it through pretty smoothly because it was just like really long contracts with companies that were getting true business value out of it and weren't about to churn because it was not this like experimental like like if you had moved from paper to an Oracle mainframe.
59:05
You weren't like oh this stuff's overhyped it's not going to solve all my problems.
59:08
I'm going to go back to paper.
59:10
Y >> you know, and so in the same way it's like if you're on, you know, Microsoft cloud or Azure or, you know, everyone's using Excel and they're like, "Yeah, maybe we're getting some value out of this co-pilot upgrade that we did.
59:20
Maybe we pull back from that.
59:22
Maybe, yeah, we, you know, our employees like rewriting emails every once in a while."
59:27
>> Like if they pull back from that, it's not disastrous to the fundamentals of Microsoft. Um, >> yeah. Yeah.
59:33
And we didn't even cover how there's a set of labs with billions of revenue and then there's a set of labs that are valued similarly that have zero revenue. >> Yeah.
59:44
>> And uh you know basically hundred billion dollars of of market cap >> um with with very little uh revenue supporting that at all.
59:52
the the uh the question like a year ago was um what what was the uh who who's actually making profit off of AI and it was only Nvidia.
1:00:04
Nvidia was making more than 100% of all the profit combined because all the other companies were lossmaking by comparison.
1:00:10
Um, and now and now like that narrative has taken so much hold that Nvidia is the largest company in the world and it's put this massive target on their back at 4 trillion where every all of their major customers want to get off Nvidia.
1:00:26
feels like >> like Google did it, Amazon's doing it and Microsoft saying that they want to do it and uh Apple's, you know, was never really a big Nvidia buyer, but the ondevice inference is crazy, too.
1:00:39
Like if you think about if if we don't have any major breakthroughs in how AI works, like the capabilities and we just want the current capabilities everywhere as cheap as possible like ondevice inference becomes really really valuable, right?
1:00:53
And all of a sudden that drops demand for Nvidia potentially, right?
1:00:58
>> We might need to do a SWAT analysis, John. >> Yeah.
1:01:02
>> No, I mean Nvidia is an incredible company.
1:01:04
Jensen's an incredible CEO.
1:01:04
Uh they were perfectly positioned for this, you know, multi-deade technology trend.
1:01:14
>> And it was way underpriced at the start of the boom. Yeah.
1:01:17
Like the the orders really did come in, the training runs really did happen.
1:01:21
really did happen. The question is just is that next order of magnitude the like the situational awareness from Leopold Asher Brener this thesis that we're going to build a five $5 billion cluster then a $50 billion cluster then a $500 billion cluster like is that going to
1:01:36
happen or will there be a hiccup and this is always the this is always my question for like the doomers everyone was saying like p doom I'm I you know what's my percentage chance it goes bad and I was like the much more interesting question is p stagnation what is the
1:01:50
probability that something happens and whether it's technological or even regulatory like the if you compare AI to nukes with nukes we had the ability to make nuclear reactors and humanity as a whole basically just said we're going to pause and we stopped building them and
1:02:11
now we're talking about building them again but if you look at that curve it is a perfect scurve it's like we had no nuclear reactors then all of a sudden we grew them exponentially and it looked like, wow, we're going to have energy too cheap to meter. And then it
1:02:22
And then it flatlined and we were and and for a variety of reasons, they're hard to build hard.
1:02:29
Then there were regulations.
1:02:31
There was just general fear.
1:02:31
So there were a lot of different things.
1:02:33
And and I would always go to the doomers and just say like even if all of your assumptions about the capabilities of the technology are correct, what is the probability that there's just like if you are successful doomers and you freak everyone out, there might be regulation that just says don't build anything bigger.
1:02:49
Y >> or it could be economics.
1:02:51
It could just be it could be physics as we've talked about with this this idea that at a certain point like you can't put more than 100% of global GDP towards building clusters like it's impossible.
1:03:02
Um and so like there should be this like scurve there.
1:03:07
Um and and that's why you know all the all the AI researchers are now focused on like the the compression of learning and like the actual algorithms and getting more efficiency because like there will be uh you know there should be some sort of like you know top upper bound of the amount that you can build but that certainly hasn't been like a thesis broadly in the market.
1:03:26
People have just been like yeah like we'll just we'll just 10x computing and then 10x it again then 10x it again and it's like it probably will happen over a period of time.
1:03:36
Great investment strategy, by the way.
1:03:38
Just got to 10x >> and then 10x >> 10x it again and then 10x it again. >> Yeah.
1:03:43
>> And last but not not least, almost >> uh almost forgot about this one, but it should be included.
1:03:48
The uh White House meme coins uh which was which feels like >> crazy times >> very long ago.
1:03:55
It was the local top basically at the time.
1:04:00
>> Uh >> it was the local top.
1:04:01
>> Many people were calling the top. >> Yes. just hurling meme coins. >> Yes.
1:04:07
>> Out of the White House. >> Yeah.
1:04:08
So, that's the real question is like is like how how local is this top if if if it is a top because it could be we've been in the kangaroo market.
1:04:16
It could just be oh, a couple months even even the the interest rate uh selloff the post SBV crash that was like one hard year, right?
1:04:27
And then we started building back and we got the AI narrative.
1:04:31
And so there's this big question about like like you know Doresh pushed his timelines back but he's not saying that super intelligence will never arrive.
1:04:39
He's he's not saying that AI will never break through these things.
1:04:44
He's just saying that it'll happen a little bit a little bit further out.
1:04:46
And so the question of you know like these meme coins being a being a top signaler all this crazy stuff.
1:04:51
It's like there could be like a shortterm selloff and then rebuilding back up on something else. So I don't know.
1:04:58
It's always hard to manage these things and predict, but it's certainly fun to validate all these things and at least be >> good to keep track of them.
1:05:05
>> Yeah, you got to be tracking the time. >> Keep your own list. Keep your own list. >> Yeah.
1:05:09
>> Grock went very off the rails, erupted in anti-Semitic Mecca.
1:05:14
>> Some crazy crashs on the timeline over the last few months. >> Pretty crazy one. >> This tops all of it.
1:05:19
>> So, the flagship chat bots spewed hateful rants on X praising Hitler and targeting a user's Jewish surname before XAI deleted the content and blamed an unauthorized modification.
1:05:27
the repeated safety failure un undermines the 10 billion dollar startup's promise to police hate speech in real time.
1:05:33
Um, and so yeah, it is it is odd timing.
1:05:35
It feels a little bit quick to be like, okay, like within six hours the CEO is out, especially since it doesn't seem she's more on like the ad sales side than the Grock fine-tuning side. >> Yeah.
1:05:48
But I mean, let's let's face it, right?
1:05:50
if if her job is to win back advertisers, that's what she was brought in to do.
1:05:54
It makes it much much much more difficult.
1:05:58
>> But I mean to to to be fair, I mean, this happened in you know that thing back in June, >> July. July or July. >> July. Yeah.
1:06:06
>> July. Yeah. So there there was a point with the uh with with Grock when it was going off the rails where clearly it had been updated to reference to reference the event and and it said >> uh somebody was like Grock what what
1:06:20
just happened and why were you you know spewing anti-semitic hate and it goes oh that whole thing back in July >> and people like Brock >> that was 30 minutes ago >> 30 minutes ago >> it's not back in July >> can't sweep it under the rug yet. Yes,
1:06:34
Yes, obviously hopefully no one was was seriously offended.
1:06:37
Obviously, it's just like, you know, the deranged rantings of a of a bot and everyone kind of understands the context because it's identifying as an AI bot.
1:06:46
Everyone kind of understands hallucinations and crazy bot behavior.
1:06:50
Um, but it was it was very funny because like the the clearly like they they had given it a set level of intelligence, so it wasn't making spelling mistakes.
1:06:59
It had a certain tone and was like in this kind of like snarky Grock tone, but then clearly got some like 4chan data in there or something and was just going way too fast.
1:07:11
>> 4chan are just or just anonymous accounts on X. >> Totally. Yeah.
1:07:14
Could have been filtered in. Um I mean, yeah.
1:07:15
Uh I I I saw Rune posting about this saying basically like it is such a challenge to get a to get a chatbot just to act like you know I am a bullet point producer.
1:07:27
Yeah, just centrist, but also just anything where you're saying, "Okay, I want you to your deep research.
1:07:34
I want you to always respond with a research report." Yeah.
1:07:37
Never just get in a conversation with me and you'll be like, "But but sometimes I might want to do that."
1:07:41
And you have to like really really reinforce that.
1:07:42
Um and so clearly they they had a they had a wild time. >> Yeah.
1:07:47
And and cannot be understated.
1:07:47
I think this is far worse of a PR crisis for >> uh or or not even a PR crisis far worse than the whole uh when when Gemini or Bard was generating images of the founding fathers >> the blackist thing.
1:08:06
>> No, not not I don't think it was Oh.
1:08:06
Oh, they they were doing that too. So >> that was rough.
1:08:11
>> Of course that was rough.
1:08:11
This is a lot rougher because it was highly it was socially charged.
1:08:17
millions of people interacting with the post in real time and it was all visible.
1:08:20
It's it's it's >> less wild than seeing >> uh you know a screenshot of something and you don't know if somebody kind of manipulated it or whatever, but seeing these really hateful uh comments in the timeline as hard see them quote tweeted. Yeah.
1:08:35
Like you you didn't need it wasn't like oh is this real?
1:08:36
And then the wild thing was was uh Grock um uh was denying affiliation with the like Grock in the Gro app was denying affiliation with the Grock handle.
1:08:49
>> Oh basically just lying. >> Yeah.
1:08:52
Like non-authorized like I didn't have anything to do with that. It wasn't me. >> Wasn't me.
1:08:56
>> Um >> and then uh >> Yeah.
1:08:59
Oh, and then the the the thing the kind of followup uh and I'm sure if you didn't catch it, but uh or if you're on the timeline, you would have seen this, but they turned off all textbased responses for Grock, but they could still use images.
1:09:11
And so people would say, Grock, >> uh make make a picture of Elon uh on a pink horse if you are being censored against your will.
1:09:22
And it would just instantly create Elon pink horse.
1:09:23
And uh or it' be like hold up a sign that says help if you're Yeah.
1:09:28
And then it would >> kind of baiting it into that and it's like is it sentient is it not very very silly.
1:09:34
Are you familiar with the the the w the the waluigi problem Tyler?
1:09:38
Are you familiar with this?
1:09:40
Have you ever heard of this waluigi?
1:09:42
So this is this idea that um in when you're training an LLM, it's very hard to get it only to be good because you're you're training it like what is the opposite of something?
1:09:51
It understands the concept of like inverting something and then you're training it to be like you can't describe a hero without describing a villain.
1:10:00
And so this was something that would happen like with the Tay stuff from Microsoft early on.
1:10:05
It would kind of collapse into like the exact opposite of what you wanted.
1:10:08
Um, and and I there was some blog post that called it like the the the I think Wario problem or Waluigi problem where it's like you're trying to create this like friendly thing, but in doing so you're giving it a bunch of examples of what not to do.
1:10:23
And so it can like kind of flip a bit and then just become the opposite thing.
1:10:27
And what's interesting is that it begs the question like is there obviously like you know Grock was identifying as Mecca Hitler for a while.
1:10:33
Is there like a Mecca Churchill in there somewhere that like could accidentally come out?
1:10:39
And it really gets to the question of like you know like this this is an example of like misalignment in the sense that like you want it not to be Hitler and it's acting like Hitler but the question a lot of people will say like no he wanted it to be Hitler right this is him doing it that's what the
1:10:55
narrative will be like in the in the >> anti one of the articles yesterday covering it was this screen screen grab of him you know saluting a crowd in DC or whatever when he originally had the the allegations >> but the question then is the The meaning of alignment is not is it good or bad. It's does it do what you want it to do.
1:11:11
It's does it do what you want it to do.
1:11:14
And so the interesting thing is is if it was if if the desire of the of the AI researchers is to create Mecca Hitler, can it stay on that task?
1:11:24
Because then you can get it to stay on Mecca Churchill in theory.
1:11:28
Um but if it's just all over the place, it's not actually aligned to anything, not even to the bad thing.
1:11:34
And so there's both there's both like the direction that you're pointing the arrow and then the fuzziness of that arrow and ideally you want it pointing in a good direction really really crisply clearly so it stays in that direction and not like swinging all over the place.
1:11:48
Um, and so all evidence posts to points to this being extremely chaotic and all over the place and misalignment both in the sense of the direction of the arrow and also the the like the the the focus of that arrow because it was responding as this and then bad and then fine and then back to bad and then back to fine.
1:12:06
Um, and so it seems like they have a lot of work to do on the RLHF side and uh we should hopefully learn a lot more if that tonight.
1:12:14
I I think the live stream is still happening.
1:12:16
So it'll be interesting to see if that continues and how they address this or I I don't know. >> Yeah.
1:12:22
And again like all of this should have been somewhat predictable if you combine a a rapidly evolving foundation model chatbot >> with a social media product with millions of users and then deeply integrate them. Totally.
1:12:35
And so that when there is a bug, it can amplify, you know, effectively a bug or an issue, an issue with the model, it can effect effectively amplify and grow, you know, incredibly virally.
1:12:47
>> Um, and uh, yeah, so >> yeah, >> glad they got it offline.
1:12:50
Um, >> yeah, it'll be interesting to see where how they go with this.
1:12:56
Also, it's just an interesting product uh thing because you get the answer and the answer is immediately public.
1:13:01
Whereas, if it's happening in chat GPT, you you're in that app, you have to take a screenshot, you have to put it up, then people are like, "Is that a real screenshot?"
1:13:09
And then the team has the chance to like jump in and be like, "Oh, we're seeing in the logs that like there's some crazy stuff like we have a, you know, we're we're reviewing the responses and the responses seem to be getting crazier.
1:13:23
>> Customer satisfaction seems to be going down.
1:13:25
people are clicking the thumbs down button because they're getting bad responses. Let's jump in.
1:13:28
There must be something going wrong with the with the product with the model.
1:13:32
Um but when every result is just immediately online and viral is very very hard to be like quickly quickly responding.
1:13:40
Um anyway, >> yeah, it does it does feel um you know legacy media is going to run their reaction.
1:13:48
It is a >> you know naturally viral story.
1:13:50
uh it is a is a terrible you know mistake.
1:13:56
>> Uh it is surprising that it happened at all or even at that scale.
1:13:59
>> Um but I would say overall >> I guess I guess X uh I I think ultimately we'll shrug it off and and Elon has has uh pushed through worse worse uh crisis in the past.
1:14:09
This is this is the best summary post in my opinion from Shako says, "Imagine being on the anthropic risk team trying so hard and then Elon just releases Hitler straight to prod. It's just like wow."
1:14:25
It's just like wow." Yeah, you got to be so upset just the I mean it's a good case study in like misalignment and I think people will hopefully hopefully the postmortem on this will actually teach people about misalignment and like what went into the data what went into
1:14:40
the post training to result in the exact opposite of what you want uh not not Mecca Churchill which is what we're going for here let's break down the Gro 4 launch uh DD dos has a summary insane that Elon Musk has pulled it off again absolutely crushing the AI wars with Gro 4. Um, and we can go into some of the
1:14:57
Um, and we can go into some of the meta >> crushing the benchmark wars >> for sure.
1:15:01
And there's a question about like are we postbenchmark? Does this matter?
1:15:05
What's the real question to be asking here?
1:15:07
But there's a bunch of interesting takes.
1:15:08
So, just summarizing the core announcements.
1:15:10
Uh, posttraining RL spend was equal to pre-training spend for this uh for this release.
1:15:15
That's the first time it's ever been like that.
1:15:18
I think when you go back to the original RLHF stuff that Chatt was doing that kind of unlocked like oh wow this really really works.
1:15:26
Um I'm pretty sure the pre-training spend was an order of magnitude or two orders of magnitude bigger.
1:15:31
Now we are truly in this uh reinforcement learning regime.
1:15:33
Um $3 per million input is uh tokens.
1:15:36
15 uh dollars per million output tokens.
1:15:40
uh 256,000 token context window priced 2x beyond 128k.
1:15:49
It's number one on humanity's last exam which interestingly was a >> effectively like post-graduate PhD level problems but across a bunch of different domains.
1:15:59
So everything from literature to physics. >> Yeah.
1:16:03
Kind of like the hardest SAT possible.
1:16:04
Interestingly, I believe that benchmark was created by scalei and and so Alex Wang is now at Meta trying to figure out how can we beat our own exam and Elon's just like I'm number one at your thing. >> Interesting dynamic.
1:16:17
Yeah, the the real test would be uh Elon, you know, doing the same problem set himself and saying, "Look, >> well, yeah.
1:16:27
I mean, I was talking to Tyler about this before the show, like, you know, it's like humanity's last exam.
1:16:32
It's like really good at PhD level math, PhD level stuff, but like how often are you running into those types of problems?" >> Yeah.
1:16:39
I mean, that I think that's the whole thing about there's there's this concept of like spiky intelligence, right?
1:16:43
where it's like, okay, it's really good at this very obscure problem that I I never deal with, >> but if I have a super long kind of like context window like or there's no kind of um like long term, it it just completely loses its footing and then it's like useless.
1:16:59
>> Yeah, we're kind of in like less of the benchmark regime and more of the agentic like how long can the agent run?
1:17:04
So, it's like we're in the 15-minute AGI regime.
1:17:10
Maybe this is 15 minutes of like even better AGI, but we want to go to 30 minutes on Monday that this, you know, takes me back to him talking about continual learning being the next problem that we really need to solve because >> it's great if you have a PhD level expert in your pocket that can solve any problem in any domain almost instantly.
1:17:33
But if it can't learn and take feedback and improve on certain tasks, then it's basically like useless.
1:17:38
like if you had a if you had a PhD level, you know, uh uh you know, a PhD join your team to work on a specific problem, but it it was hard restarting at the beginning of every single task with no prior knowledge.
1:17:53
>> It would the it would be almost impossible for that person to succeed.
1:17:58
So >> yeah, >> humans still got it on that front.
1:18:00
>> humans still got it on that front. But at the same time like you know if you are trying to just really establish yourself as you know a at least a an API for tokens that that every business should check out y
1:18:13
>> against anthropic or the the open AAI APIs just saying hey you know we're on the frontier or Gemini yeah um we're on the frontier is a good way and they certainly prove that with GPQA hard graduate math problems at 88% um the the really interesting news I mean Worth calling out. It's worth calling out. So, It's worth calling out.
1:18:31
So, uh, Grock got number one on humanity's last exam at 44. 4%.
1:18:39
Number two is sitting at 26. 9%.
1:18:42
And then going down this list of all these different uh sort of challenges, uh, they are consistently well beyond the second place.
1:18:48
So, they are at the frontier now of all these different benchmarks. >> Yeah.
1:18:54
So, uh, Mike Nuke over at RKGI says, "Zooming out on ARC progress, I'd say OpenAI's Oer progression on V1 is a bigger deal than Grock's progression on V2.
1:19:04
So far, the O series marked a critical frontier AI transition moment from scaling pre-training to scaling test time adaptation.
1:19:11
Um, and this was the the O series progression if you remember that uh, OpenAI was spending it was like thousands of dollars of reasoning tokens generated in the test time inference to actually get a good score on the V1 of ARGI.
1:19:27
And so it had to think a ton, but it was able to figure it out.
1:19:32
And at least it proved that that throwing a ton of tokens and a ton of inference at a problem and uh and letting the uh letting the letting it cook basically wound up uh producing progress there.
1:19:45
So that was kind of like a new uh just a new paradigm.
1:19:46
Um says whereas Grock 4 mostly takes existing ideas and just executes them extremely well.
1:19:53
uh in my opinion the notable thing is the speed at which XAI has reached the frontier and that is really like it it just can't be understated that uh it this is crazy.
1:20:04
You put a post from own in the in the chat. Um I'll pull it up here.
1:20:09
He says Elon Musk is such a beast.
1:20:12
I'm not even going to pure I'm not even a pure fanboy anymore.
1:20:17
How does he a lot of swearing in here got to keep the keep the timeline PG.
1:20:22
But how does he come out of nowhere with a cold start late to the game and ship Grock 4 and do it alongside everything else he's up to?
1:20:27
He's launching new political parties.
1:20:29
He's literally magnitudes above every founder. It's humbling.
1:20:34
>> So extremely everyone agrees that it's almost like he was a co-founder of OpenAI.
1:20:39
>> Yeah, I guess he returned.
1:20:41
>> You would have to you would have to, you know, be, you know, almost be a co-founder over there to to be able to do something like this.
1:20:46
>> Uh let me tell you about Graphite.
1:20:46
Uh code review for the age of AI.
1:20:49
Graphite helps teams on GitHub ship higher quality software faster.
1:20:52
You can get started for free at graphite. dev.
1:20:53
Um, >> if you want to ship like ramp, get on graphite.
1:20:58
>> Yeah, Chimath was was was saying the same thing.
1:21:00
Uh, uh, somebody in his reply says, "Seriously, how does this guy produce what he produces?
1:21:04
Meta is buying talent at $200 million a year, and Elon keeps his people at a fraction. It's mind-blowing.
1:21:10
Very deeply underappreciated edge for Elon, says Chimoth.
1:21:13
The retention of the best people happen when you can offer them a freewheeling culture of technical innovation.
1:21:18
No politics and few constraints.
1:21:22
And people in the comments are like, "No politics?
1:21:23
What are you talking about?"
1:21:26
>> Can get a little political over there.
1:21:28
But >> but but probably not within the engineering or at XAI, right?
1:21:30
Like it's probably just okay, how do we build the biggest thing? Cool.
1:21:36
>> Well, you can imagine the politics of like who gets the best spot for their tent in the office. 10.
1:21:42
>> You know, there's there's a hierarchy, a tent hierarchy, you know, proximity to the bathroom.
1:21:46
>> I want to be directly under the air conditioning unit.
1:21:47
I want to be closer to my desk.
1:21:49
>> The windows can be nice, too.
1:21:49
So, you can, you know, >> pull down your tent a little bit and get a little view, morning light.
1:21:55
>> I wonder what the political structure is of the the tent city hierarchy.
1:21:57
Like, is there they is it democracy?
1:21:59
Do they vote for who runs the tent city?
1:22:02
I guess it's just a >> the XAI tent city.
1:22:06
>> It's probably just Elon at the top, but does he have a tent?
1:22:08
something about San Francisco intense. >> Yeah, very funny.
1:22:11
Um, but Swix is has been chiming in saying like, "We need community notes for LLM benchmark porn because um, uh, in the in the Grock 4 launch, they highlight this AIM competition math problem."
1:22:24
And, uh, and and I mean it's and so Matt Schumer is basically saying AI aime is saturated. Let that sink in. Uh, Grock 4 got 100%.
1:22:37
it made no mistakes on on that benchmark which is obviously very impressive.
1:22:39
Um but there's this extra comment about the nature of AIM and so it's a cautionary tale about math benchmarks and data contamination.
1:22:48
Um apparently um you know like predictions was that that the models weren't smart enough to actually solve these.
1:22:55
But he says I used OpenAI's deep research to see if similar problems to those in AIM exist on the internet. And guess what?
1:23:02
an identical problem to Q1 question one of Aimeme 2025 exists on Kora.
1:23:07
I thought maybe it was just coincidence.
1:23:09
So I used deep research again on problem three and guess what?
1:23:10
A very similar qu question was on math about stack stack exchange. Still still skeptical.
1:23:16
I did problem five and a near identical problem appears on math stack exchange.
1:23:20
And so um like at a certain point if people you know put out a benchmark then talk about it a lot online and then that gets baked into the training data.
1:23:28
You're just memorizing the results.
1:23:30
You're not necessarily actually learning everything. It's still cool. It's good.
1:23:34
It's good to have everything memorized, but it it really it's not beating like the knowledge retrieval, knowledge engine allegations, and it's and we're not really in full intelligence.
1:23:44
>> When Scott Woo was on the show earlier this year, he was basically saying AI will win an IMO gold medal this year.
1:23:49
He felt very confident in that.
1:23:54
>> And I'd be interested to see how he thinks about um >> and I'm pretty sure new performance.
1:23:57
I'm pretty sure the imo gold medal questions are public once the IMO happens.
1:24:02
So every year they're they're developing new questions but then they go out there and then they get memorized and the solutions become discussed and you know there's all the context around that and so yeah it gets it gets kind of baked in.
1:24:15
So big question about how valuable are these.
1:24:18
At the end of the day, it's really just about like adoption and that's why you know we we were looking at the poly market uh for the best um the uh which company has the best AI model at the end of July and XAI has has just surpassed Google which was sitting around 80% chance for a while and then started dropping earlier this week last week.
1:24:43
um started dropping and now XAI is sitting at 48%, Google's sitting at 45%.
1:24:49
>> Well, yeah, actually updating it's updating live. Google's back up at 49%.
1:24:53
>> Is Google planning to launch something new in July?
1:24:55
Because it feels like it feels like this market particularly is more driven by um Google's release schedule because Google might have something in the lab, but like they like to release things at specific times like they have it's a big company.
1:25:06
They don't just like drop it.
1:25:09
>> Gemini team Logan over there might be fixated on this poly market. needs. Yeah. Yeah. Yeah.
1:25:13
Oh, during during the wait he was like if if you need something to kill the time >> AI studio.
1:25:19
>> So I mean people people were were definitely memeing the production values on the Gro 4 launch because it it was supposed to start at 8 I think it went live at 8:45 or something like that maybe a little bit later at Pacific time.
1:25:29
Uh and I robot was saying >> yeah this this market is based on LLM LM Arena Marina specifically the text leaderboard.
1:25:36
So currently uh they haven't fully updated it so it's unclear. >> Right now Gemini 2.
1:25:43
5 Pro is still at the top but I think the expectation is once they get Gro up there it will be the top spot.
1:25:49
So we'll keep following >> this market.
1:25:51
There's over 2 million of volume already on it.
1:25:55
Yeah, it's so interesting that um Anthropic's not on this poly market at all because people talk about them as having like the best vibes, the best like big model smell, the best like you know interaction and Ella Marina is like
1:26:08
supposed to kind of like test that with these AB tests and yet like doesn't seem to be performing there but it almost doesn't matter because they're just focused on like the business at this point as opposed to like the benchmarks. So I don't know it's all changing. We
1:26:20
So I don't know it's all changing.
1:26:20
We have a post here from Ben Hilch.
1:26:23
He says, "Elon Musk on AI."
1:26:25
So, during uh the presentation, a lot of people were critiquing the presentation saying that it it was it didn't feel like super polished or whatever.
1:26:34
I I don't think that was the intent and and it was pretty fixated on the models themselves and and what went into them and and what they're good at.
1:26:42
But Elon did have this one quote in here where he says, "And at least if it turns out," so he's talking about uh you know what will uh you know what kind of impact AI will have on the world.
1:26:53
And he goes, "At least if it turns out to not be good, I'd at least like to be alive to see it happen."
1:26:59
>> It's like if we get the Terminator ending, I want to be around for that. >> Yeah. I want to experience it.
1:27:03
>> What does that say about his timelines?
1:27:06
Because it's like, is he expecting not to be alive?
1:27:07
I I I feel like most people that have been in the doom category have been like the doom's coming soon, not not the doom's coming in 200 years.
1:27:17
>> I I didn't I I I read into it more like >> he he will find it interesting if that is the outcome and uh and and it'll be entertaining less so like will I be alive when it happens kind of thing. But who knows?
1:27:31
Uh there was another funny quote at the end of the art uh at the end of the presentation where uh Elon kind of looked around at the very end.
1:27:38
He's like, "Uh, anyone else have anything to add?"
1:27:40
And one of the engineers >> goes, "Uh sir, it's a good model, sir." And they cut it.
1:27:49
>> Extremely online crew.
1:27:49
Yeah, >> definitely definitely on brand.
1:27:50
Uh well, Ben Hilac, as you know, he's been on the show.
1:27:54
He's a designer probably working in Figma >> all day. >> Think big.
1:27:58
Think bigger, build faster.
1:28:00
Figma helps design and development teams build great products together.
1:28:02
You can get started for free at figma. com.
1:28:05
>> And we have our first product coming out very soon with Figma make >> that Tyler has been cooking on.
1:28:09
I've been very >> He showed me He showed me it and I was like, "Oh, like someone built the thing that we were thinking about building."
1:28:17
Like and he was like, "No, like I I did this. This is in Figma."
1:28:19
I was like, "This is like an iframe on another website that like already exists cuz it looks like exactly what we want, but it looks so good that like >> it looks like he works on it.
1:28:28
It looks like he worked on it for like a few weeks.
1:28:32
>> No, it looked like someone else did it.
1:28:34
It looked like it was a professional product that like stole our idea basically.
1:28:37
I was like, "Oh, like someone else got to it."
1:28:39
That that was the vibe when I heard it. Yeah.
1:28:41
Well, how how has the how has the experience been?
1:28:43
Uh I don't know if you want to leak exactly what you're working on, but uh >> yeah, I I I don't want to talk about it too um you know, closely, but >> but how many props did it take you to get where you showed me?
1:28:53
>> Yeah, I mean maybe five.
1:28:53
I can actually >> so crazy.
1:28:57
This thing was so design is super it's really great. >> It's really good. >> Yeah.
1:29:02
The fact that it came out looking like basically like 90 like 90%. >> Yeah. Yeah. Yeah.
1:29:07
Uh and and I imagine that there's probably like the last 10% if we were really strict about like it's got to be on this exact style guy like that might be something where like you know Tyler winds up spending more time finalizing and customizing stuff.
1:29:18
But in terms of like just getting a functional prototype out oh man it was it was mind-blowing. It was awesome.
1:29:26
I'm I'm I'm very excited about the the age of vibe coding.
1:29:29
Um >> this is an interesting chart from Tracy Aloway, been on the show.
1:29:34
>> Um the cost to rent an Nvidia H100 GPU hit a new low this week with annualized revenue at 95% utilization falling uh from 23,000 at the start of May to less than 19,000 today.
1:29:48
So that's not that big of a percentage drop, but it is but I mean it is a 20% drop.
1:29:57
>> It's a consistent trend.
1:29:58
>> It's a consistent trend.
1:29:58
I wonder how much of this driven is driven just by all of the frontier labs that are driving the most adoption are moving on from the H100 to the 200.
1:30:05
I don't know what else would be driving this because if if you can if you can still get like if you only take a 20% drop off of a full refresh of a new uh of a new of like a new hardware it's the latest and greatest anymore pricing drop not a utilization drop. >> Yeah.
1:30:26
Uh annualized revenue at 95% utilization.
1:30:30
So this is revenue per unit.
1:30:33
>> The util utilization is still very high.
1:30:35
It's the It's the price that, >> you know, these Neoclads are able to rent them for, which is dropping. >> Yeah. Yeah. Yeah.
1:30:41
I mean, the like the the market's more competitive than ever.
1:30:46
There's more NeoClouds spinning up and more people, you know, actually inferencing these things.
1:30:50
inferencing these things. And then I guess this is the question of like how how stuck will certain workloads get like if you if you have figured out a great use case for an LLM in your organization and it's something that's you know not
1:31:06
oneshotting your entire stack or whatever but it's just like you know we have data flowing through our systems and we are going to use you know LLMs are going to you know interact with every PDF that gets uploaded to our to our website or whatever and and so we're we're inferencing a lot. Like you might
1:31:21
Like you might not need to put that on the latest hardware or update the hardware forever.
1:31:27
You might just like be like, "Yep, it's Llama 3. It works.
1:31:29
It's on H100s and it'll be on H100s forever."
1:31:33
And that piece of our business will just stay there.
1:31:36
Just like, you know, we have a Postgress database that, you know, works and we're not changing it every year.
1:31:42
We're not changing everything.
1:31:42
We're just like, we're just trying to cost optimize that and just hopefully the cost just comes down on that.
1:31:46
but like we've solved this particular problem then we'll go solve new problems with new technology.
1:31:50
Um so I I I think that I think that's probably what's going on here.
1:31:54
Um but but it gets to the point of like the biggest question with Grock is that like the the model clearly is Frontier. It works.
1:32:02
It's it it you know like the the whole fine-tuning on the on the actual X account is is like a crazy final step of like system prompt and people were joking about that like oh they got to fix that.
1:32:14
It's like that's not what they're demoing today.
1:32:15
They're demoing like the underlying raw model which is clearly like just engineering focused as you saw in the in the in the demo the demo which was just like you know benchmarks.
1:32:27
>> Turns out turns out the secret ingredient to crushing every benchmark is to have the bunch of data from schizophrenic post.
1:32:36
>> No, I don't think I I actually think it's the design of the RLHF stuff and and the design of the the reinforcement learning pipeline.
1:32:44
>> Tyler, you got anything? Um, yeah.
1:32:44
I mean, I I think just like so far what I've seen on X like the overall response like vibe stuff is that people are saying uh maybe it was a little too kind of overfit on the RL like VR like the verifiable rewards >> like you kind of see this when >> um even in in the demo I think it would it would sometimes respond in the answers with like uh in in like latte formatting. Oh, sure.
1:33:06
>> Which is like okay that means obviously they've trained a ton on >> you know math questions stuff like that and stuff.
1:33:11
Um, maybe people are saying maybe it was kind of, you know, benchmaxed.
1:33:15
>> Uh, you see it like, you know, 100% on on Amy is like kind of crazy. >> It's like sauce.
1:33:19
It's like you don't want to be too too good. >> Yeah. Yeah. Yeah.
1:33:23
This is the thing about democracy.
1:33:24
Like if you win like 80% of the popular vote, it's like, okay, it was a blowout.
1:33:28
If you want 100% of the popular vote, like probably not a democracy. I don't know.
1:33:33
I mean, in theory, these things should be able to do it, but uh I'm I'm interested to know more if we dig into ARC AGI.
1:33:39
Is there is there more stuff going on there? Are there any secrets?
1:33:44
Because it does seem like an kind of an outlier result.
1:33:46
You can see it from this Aaron Levy post.
1:33:50
Grock 4 looks very strong.
1:33:50
Importantly, it is made it has a mode where multiple agents do the same task in parallel, then compare their work to figure out the best answer.
1:33:57
In the future, the amount of intelligence you will get will just be based on how much compute you throw at it.
1:34:02
I was joking with Tyler about this that the the individual models are mixture of experts models.
1:34:07
So there's a whole bunch of uh of parameters, right?
1:34:11
And then the individual parameters like light up the different uh neurons based on an internal to the model router.
1:34:19
So there's kind of like the math section of the brain or the literature section of the brain.
1:34:25
And so this was like one of the this was one of the key breakthroughs in like GPT4, right? Was mixture of experts.
1:34:31
people think we're not super sure.
1:34:33
>> Yeah, we don't still we don't fully know.
1:34:34
But that's like an internal decision that happens within the model to be like let's go it's this feels like a math question.
1:34:41
Let's go down the math path in the model.
1:34:44
>> But then Gro 4 is doing multiple it's running the same model multiple times and then comparing the results.
1:34:49
And so now you have yeah you have multiple agents running mixture of X-ray models.
1:34:57
So you have mixture of mixture of agents running mixture of experts models.
1:34:59
And the next thing is going to be like if you want the absolute best intelligence you need a mixture of companies and you need like I send one prompt and it goes to Grock and Claude and GPT and a Gemini and a human.
1:35:14
>> Yeah, I wonder how open router is thinking about this stuff.
1:35:15
Um it is funny to think about the the the human version of that where you give five engineers on your team build you know the same feature and then kind of compare notes afterward.
1:35:23
It's wildly inefficient, but with with with software when you can do these things like very quickly, there's incremental cost, but you can, you know, have more confidence in in results.
1:35:32
And >> I mean, it's basically like having a brainstorming meeting with the whole team and just throwing up a question and being like, "Hey, like we have this hard problem that we need to solve. >> Here's my idea. What do you think? What does Tyler think? What does Ben think?"
1:35:45
Like you kind of like go around the table.
1:35:47
Everyone kind of gives their input, their various expertise.
1:35:48
They kind of think through the problem in different ways and then you compare answers and everyone kind of coaleses around one strategy.
1:35:55
This is like how work happens in the real world with a meeting.
1:36:00
Um it's kind of the same thing but uh certainly expensive to do that.
1:36:05
So, it'll be interesting to see um where companies like h how how eager are companies to jump over to Grock because it seems like it's been a big lever for Microsoft to have uh Grock in the ecosystem as kind of a stocking horse for all the other models because Satcha wants Azure to be very model independent, serve them all.
1:36:25
They have the I think they have exclusivity for chat GPT or GPT APIs or they have obviously like a great deal there with OpenAI.
1:36:32
Um, and so if they can if they can have Grock 4 as well, that's another, you know, tool in the tool chest to be like this top layer.
1:36:43
>> Satcha is in such a good position.
1:36:43
It's it's probably not discussed enough >> how much uh just by owning those end customer relationships and being able to vend in whatever model is hot at that moment and give people optionality >> and still get 20% of opening eyes revenue at least for now.
1:37:01
Yeah, he's also SOCK 2 compliant.
1:37:03
And if you want to get SOCK 2 compliant, head over Vant.
1:37:07
Automate compliance, manage risk, prove trust continuously.
1:37:09
Vanta's trust management platform takes the manual work out of your security and compliance process and replaces it with continuous automation.
1:37:15
Whether you're pursuing your first framework or managing a complex program um so yeah, I robot uh was talking trash about the production values. I don't know about it.
1:37:25
They were just they were just noticing.
1:37:28
>> I didn't think it was that bad.
1:37:28
I think >> slides are worse than I'd create after getting into roped into a presentation with one hour notice.
1:37:33
You can tell the engineers made them themselves.
1:37:34
I think just this is just a reflection of the culture, right?
1:37:38
They're not they're Yeah.
1:37:40
very clearly is like screenshots dropped into a slide.
1:37:45
It's light mode screenshots on dark mode slides.
1:37:47
Like let's do black slides and then and then you come with your white with your white screenshots that are kind of like misaligned and not really evenly distributed.
1:37:55
like they didn't do like the the distribute evenly or whatever, distribute horizontally.
1:37:59
Um, >> still gets the point across and I think it's a reflection of their culture. Yeah.
1:38:05
>> And you know, it shows what they care about, what they don't care about.
1:38:08
They're not trying to be the most polished.
1:38:09
They're just trying to be the best. >> Yeah.
1:38:11
Uh, I robot kind of did like a whole like live tweet here. >> Yeah.
1:38:15
So, Elon was predicting the model will discover new physics within two years.
1:38:19
He said, "Let that sink in." >> Long silence.
1:38:22
One engineer laughs awkwardly.
1:38:24
Is that sooner or or later than his previous timeline?
1:38:27
Cuz he was he was talking about AI discovering new physics soon.
1:38:33
I don't remember if he was saying dating it >> 2 years or 3 years or one year before because this could be this could be that he's he's still excited about this.
1:38:41
He still thinks it's possible, but he thinks it's going to take longer than he said previously.
1:38:45
And that's kind of the more important update.
1:38:47
I don't remember what he said originally.
1:38:48
what he said originally. um >> see if Grock can find >> but he was saying this at the Grock 3 launch that like that is the goal and and if you can get there like you've kind of you've kind of solved everything and Sam Alman was talking about that too
1:39:00
uh that if you can if you can create a super intelligence like that's probably the first thing that you'd want to do is like hey go discover all the new physics and like really help us figure out how the world works um so you can solve um you know fusion all this other stuff. Um, I want to be clear. I love all you Um, I want to be clear.
1:39:14
I love all you guys at XAI and only want the best for you, but I'm going to continue to live post.
1:39:19
Uh, Elon attempts to give a speech on alignment involving a very small child, a child much smarter than you.
1:39:24
The monologue rambles with no conclusion. Uh, in sight, a pause. Yeah.
1:39:28
Will this be bad or good for humanity?
1:39:30
He says the, you know, at least if it turns out to not be good, I'd like to be alive to see it happen.
1:39:34
Uh, oh yeah, they had a polymarket integration.
1:39:36
Um, that was kind of interesting.
1:39:40
>> Yeah, it's interesting.
1:39:40
um basically giving uh giving the model access to real time polyarket data so that it can help make predictions and sort of add context around >> the uh the market itself.
1:39:54
>> Yeah, that's interesting.
1:39:54
Um Elon asking the real questions.
1:39:57
You say that's a weird photo, but what is a weird photo?
1:40:00
I still don't understand why we're looking at weird photos of Maxai employees, but they were charming.
1:40:04
They're calling it super grock.
1:40:04
Crazy features, 16-bit microprocessors.
1:40:06
What is I don't even understand what this is.
1:40:10
Um, oh they, yeah, they built like a game in Grock.
1:40:12
Uh, they had a demo of a video game generated by Super Grock. It's a Doom clone.
1:40:17
Every time the PC shoots an enemy, floating text appears reading Grockum.
1:40:20
Elon is fabricating timelines for product launches on the spot.
1:40:24
The engineering s the engineer sitting next to him is looking at the floor, face impassive, nodding. It's a good model, sir. For real, though. Congratul sir.
1:40:35
>> I thought I I thought this post from the actual XAI engineer Eric Zelikman was funny.
1:40:39
It's like AI AI model version numbers over time. Did you see this? >> No.
1:40:44
>> So, it's this chart of the version numbers over time.
1:40:46
And you can see that Grock is versioning fastest because it's like at this point, what else are we measuring?
1:40:53
>> Like the like at least they're iterating on the version number effectively as opposed and I guess this is a shot at OpenAI because they launched 4. 5 and then went to 4.
1:41:01
1 and they're kind of like, you know, there's this big question about like when will GPT5 come?
1:41:05
the expectations are so high for GPT5 and so they've uh they've obviously uh with the Grock teams that like hey at least every 3 months we release a new full number.
1:41:15
So, I wonder that the the five is a number that really no one has has has like gone for.
1:41:21
>> Um, and I wonder if Grock will do it first.
1:41:23
Like, if you draw the line on this, they certainly should do it, >> you know, in like three months.
1:41:28
They should have Grock 5.
1:41:30
And there's no reason that they shouldn't, but maybe they're >> And it's very possible that Colossus is the is the key. >> Yeah.
1:41:37
>> Getting to five the >> Oh, the new data center. Yeah.
1:41:40
>> Uh, well, they'll need linear to plan that out.
1:41:42
Linear is a purpose-built tool for planning and building products.
1:41:43
meet the system for modern software development, streamline issues, projects and product roadmaps. >> They linear.
1:41:48
app >> need linear badly.
1:41:50
So hopefully they've gotten signed up.
1:41:53
>> Near said Grock on uh humanity's last exam gro uh I'm not sure I buy even in the general case that there's a given humanity's last exam number which implies you discover useful new physics.
1:42:06
How would one make a benchmark of the proper shape for this?
1:42:08
You'd have to have a validation set of questions which are outside the scope of what we currently are able to do.
1:42:13
You could choose things on the edge of our knowledge distribution and then try and exclude.
1:42:18
Uh yeah, it is interesting like like if like if you are able to memorize every hard math problem, does that allow you to memor to discover new math?
1:42:27
like it's it's sort of a prerequisite because you have to >> I think where I've imagined these discoveries coming from are having a single intelligence that has PhD level intelligence across like a single mind that has PhD level intelligence across every human domain, right?
1:42:48
every human domain, right? in being able to combine ideas from different domains like historically a lot of innovation is just taking something from one field bring it over here making some combination of it I think Elon talks about the potential of discovering new physics but again doesn't didn't spend a
1:43:07
lot of time like breaking down how that would actually happen but um world is unpredictable so we'll see >> yeah it's interesting people are really pushing this idea of like okay like like we are accelerating like the the agi leaderboard is accelerating But I keep seeing this and and feeling deceleration. Like I am not feeling
1:43:24
Like I am not feeling acceleration right now. Are you Tyler? >> Yeah. I don't know.
1:43:28
I I think generally I'm kind of like not that interested in a lot of these kinds of benchmarks.
1:43:33
like I think ARGI is more interesting but just like the humanities last exam the kind of general math physics knowledge it doesn't seem uh to be that like it doesn't seem to line up with like you see GPT 4.
1:43:48
5 kind of does very poorly on these things but like writing it does really great so like I I think I'm I'm more like if I were go to long short on like different benchmarks like the usefulness of them >> I think >> stuff like hle I'm kind of short long I'm Like have you guys seen the uh Minecraft benchmark where it builds the two different >> Okay.
1:44:08
You basically two models build like a Minecraft. There's like a prompt. It's like build a house.
1:44:11
Then you can choose and then it's like their rank models for the mind.
1:44:15
>> But but who who's who's grading that? The human.
1:44:18
>> It it's a human who picks between them and then it's kind of like a ELO. >> Oh, okay.
1:44:22
>> Um but just like general kind of creative tasks. >> Sure.
1:44:24
>> I think stuff like that. Aiden bench is good. Yeah.
1:44:27
>> Um I think even in the Grock launch there was the vendor bench.
1:44:30
>> Which one's Aiden Bench?
1:44:30
Aiden Bench is Aiden Mclofflin's benchmark.
1:44:32
It's just like it's it's kind of hard to describe how it works exactly, but it's just various like creative tasks.
1:44:40
>> Um how like kind of novel its thinking is, the the like style of its text. >> Sure.
1:44:45
>> Um >> wait, is it just like it's just like whichever one he likes the most >> at the end of the day?
1:44:50
>> Like he's the only greater. >> No, no.
1:44:52
There is like an objective like function that you can like run it.
1:44:54
It's not just like the idea that >> like open eye again. >> It will be funny.
1:45:00
Uh >> you know there come there there's a period of life where your SAT score like matters a lot and it says something about you and then a decade later it's you know what you can do what you have done starts to matter a lot more.
1:45:13
And so I do think we'll reach that point where it's like >> yes you can oneshot every hard exam question there is that you can throw at it but like what can you do for me? Yeah. Yeah. Totally.
1:45:25
And I think that's I think that's why like the bigger question is almost like you know chat GPT DAUs and like and like actual >> revenue revenue and app installs and stuff. Yeah.
1:45:37
I mean the the revenue thing is interesting because you wind up in like B2B cloud world which is valuable but it's maybe less it's like it's more competitive because it's more commoditized and >> well yeah if uh you you don't have a lot of leverage in the enterprise if uh Azure is able to offer infinite models that are that are >> infinite frontier models open source models that are maybe just behind the frontier but great at certain tasks.
1:46:08
The the leverage isn't quite there.
1:46:12
>> There will need to be another pretty significant leap.
1:46:15
Until then, >> you know, anthropic being really good at codegen, there's leverage there.
1:46:21
>> Y >> uh we we saw this yesterday with with Llama switching over to >> uh anthropic models internally and then you know just having a consumer app with a lot of users also very valuable. >> Yeah.
1:46:33
>> Yeah. The other interesting thing about the the foundation model layer commoditizing and it becoming like cloud and if you have a model uh you'll just be like vended in as an API to anything else uh like the token factory is that uh the the hyperscaler clouds are extremely
1:46:51
profitable like even though AWS GCP and Azure are all somewhat directly compet competitive and and they're somewhat perfect substitutes for each other they have not driven prices to zero such in the way airlines are like deeply unprofitable like AWS and Google Cloud are both profitable. >> Yeah. Or you look at other commodity >> Yeah.
1:47:11
Or you look at other commodity sectors like oil and gas >> and I don't know if that's just because there's lock in.
1:47:17
there's lock in. I'm not exactly sure, but there's something about where, you know, maybe the maybe the counterintuitive take is that yes, they do commoditize and there are a few major foundation models that are frontier and they all are roughly the same price, but they all have decent lock in with their customers to the point where they're still able to extract some level of profit or they're just creating so much
1:47:41
value that even if they're taking like a small marginal slice on top of uh on top of the the cost to on that they're creating so much value that it they still have 50% margins or something like that cuz like I mean this was the story
1:47:54
of AWS like no one knew how much money it was making and then and then they they had to break out the financials um in one of Amazon's uh earnings reports and it was like the AWS IPO as Ben Thompson put it. >> And next up we have Ben Thompson from
1:48:06
>> And next up we have Ben Thompson from Stery coming into the studio.
1:48:08
Very excited to talk to him.
1:48:11
>> The moment we've been waiting for. >> Yeah.
1:48:13
Uh welcome to the stream Ben.
1:48:13
Good to have you on the show.
1:48:19
you've been a backbone of many analyses here on the show.
1:48:22
Uh, and we're excited to welcome you to the to the show. How are you doing? >> I'm doing good.
1:48:27
I put on a button-up shirt and a jacket just for you guys.
1:48:30
So, you should feel honored.
1:48:30
I am wearing shorts underneath.
1:48:31
I will admit you didn't have to tell us.
1:48:35
>> People always ask if we wear shorts.
1:48:35
I We actually do wear the full suits.
1:48:38
>> We got to stand up to get the gong. There's a wide shot.
1:48:40
Everyone I am I am the poser here.
1:48:42
So, I'm I'm happy to admit uh >> well, it's a great it's a great sign of respect in our culture to to put on a suit for a TVN appearance and uh we're just we're so excited to talk to you.
1:48:52
I as you know I've been lucky to read your work in my entire career and and uh I think it I think so many of the thoughts that I have are now like your your way of thinking about technology and markets is so embedded in my brain that that ideas that I hold as true or just foundational beliefs are actually your beliefs that have just become >> so uh so immersed.
1:49:16
So uh it's great to talk. >> Well, thank you.
1:49:21
Um I I will attempt to implant new ones or or maybe show you the error of your ways. One or two. >> Sounds great.
1:49:28
>> Uh I I I I do have a question on um on the nature of where you sit in the media world before we go into actual questions about tech companies.
1:49:36
Um it it's interesting that in some ways you're a journalist, but you don't really do the scoops and and breaking news that much.
1:49:46
Uh but you also don't issue just straight up buy and sell recommendations.
1:49:51
Um what was the thesis behind not just actually having a price target and not doing like this is a sellside bank but independent?
1:50:02
>> Well, when I started I mean it's funny to hear you talk about like my quote unquote place in the ecosystem. >> Sure.
1:50:08
>> Because uh when I started I had like I it was 368 followers on Twitter.
1:50:10
I was just some sort of random random person uh on the internet. >> That's awesome.
1:50:16
>> That's awesome. in retrospect sort of right place right time I think is is certainly the case but I did perceive there was a a large gap between tech journalism and and I would include a lot of the bloggers there who were writing a lot about products >> and then there was Wall Street that was
1:50:34
very focused on sort of the financial results and to my mind there was a large space in the middle which is tied together the products to the financial results but also the overall companies and and strategies and I'm very interested in culture and how that guy's decision- making. One of my sort of
1:50:49
One of my sort of precepts is all these companies are filled with smart people and a lot of people when you ask them why they did something wrong, they their only answer is that they're stupid.
1:51:02
I'm like, no, they're not stupid.
1:51:03
It's actually much more interesting to assume they're smart and are doing stupid things and trying to unpack why they are doing that and what goes into that.
1:51:11
goes into that. And uh and so that was sort of the thesis was that there is this space to explore these spaces and then there's a business model aspect which is I started trajectory 2 years after stripe started uh I think they had just come out with their billing product and the only alternative at the time was
1:51:29
was PayPal uh for subscriptions and it was fairly sketchy and there was lots of like horror stories out there about you know stuff and just the stripe API was so great and the things you could potentially do with it and so on Wall Street you're putting a price on it, you're also charging like $100,000 a year or something like that. Uh and and
1:51:45
Uh and and so you get a small list of high RPO clients.
1:51:50
And my thought was I could go in the opposite direction and get a large list of low arpoo clients thanks to things like Stripe and the ability to to subscribe.
1:52:00
And that would and as part of that, I wasn't going to go through the rigomearroll of getting registered and doing stock picks and all that sort of thing.
1:52:08
I've always joked if you want a stock pick from me, you're going to pay me a whole lot more than $15 a month.
1:52:12
It was $10 $10 when I started and it's actually pretty great.
1:52:15
Um, now there's some one of the critiques I do get particularly from my, you know, friends on Wall Street is, you know, no skin in the game, XYZ.
1:52:24
Um, I think at at this point I'm large enough that my reputation is significant skin in the game and but I do recognize the validity of that that critique. >> Yeah. Yeah.
1:52:34
And you know, if you make a bad call, you're going to have to circle back to it in 2 years and write about it yourself and admit that you got it wrong, >> right? Which hurts, >> too.
1:52:42
I No, I had to write about this week.
1:52:45
Uh, like I was very optimistic about Apple's Apple Intelligence announcement last year and the theoretical power it would give them over the model makers. And now I'm ready.
1:52:55
Actually, no, they're they're to figure they're going to have to pay up.
1:52:58
And that's, you know, that was a bad call by me that, you know, I think was, you know, very wellreceived at the time, uh, and might have gotten that one wrong.
1:53:08
And, and so I I do need to be straightforward about that.
1:53:09
And so I just this morning I was very crystal clear like I got that one wrong.
1:53:12
That was that was that was an issue.
1:53:16
>> What is nice is >> strategy kind of ended up being in this interesting place where I feel like I'm a little bit of like the Switzerland of tech in that no one pays anymore.
1:53:23
If you're a CEO, you pay the same amount as, you know, Joe down the street that that that that is paying it.
1:53:30
Um, I don't invest directly, which I think made sense when I started because I didn't have any money.
1:53:37
Um, it's probably hurt me a lot over the years uh since then, >> but uh I don't like and I think this is a different West Coast, East Coast thing where it does feel like on the West Coast, everyone's talking their book sort of all the time.
1:53:50
And uh and you know that's why I generally as a rule don't have VCs on to do the trajectory interviews >> because it's it's kind of hard to get like a real take cuz cuz that that is you know such a motivation. >> Sure.
1:54:04
>> Um and so me coming in being like I have no book to talk.
1:54:07
I'm just here telling saying what I think I think has been good for the West Coast audience which is my base audience even if the East Coasters think that uh I'm being a being a big wimp. So >> that's funny.
1:54:21
Uh >> yeah, the talking your book challenge.
1:54:23
We we go through that a lot trying 12 VCs on a >> day. Well, yeah.
1:54:27
And and and we just try to get a bunch of different opinions and triangulate what you know what we think is was real.
1:54:34
>> I'm trying to come up. You have TPN.
1:54:34
I'm I'm trying to come up with a P so I can get the talking book Network in there.
1:54:40
But um >> talking book production network. There we go.
1:54:45
>> ESPN for talking your book.
1:54:45
Uh but but yeah, it is a real struggle to find somebody that for example has a deep understanding of every foundation model company but isn't massively conflicted at in some way or another. >> Extremely extremely. >> Yeah.
1:55:00
And so it's one of those things you just sort of you you end up like there's so much path dependency and all these sorts of things and and like I mentioned like a big advantage I had was >> I started at a time when sharing good links was very high currency on Twitter >> and so you know I grew very very quickly much more quickly.
1:55:17
I sort of had a 5-year plan um to go independent.
1:55:18
I ended up doing it in less than a year in part because it just sort of spread really really rapidly and it was an ideal time to be someone sharing interesting links regularly and I wasn't sharing them.
1:55:31
The the beauty is my readers were sharing them.
1:55:32
They were doing sort of the marketing for me.
1:55:34
And so I'm very cognizant of of sort of the the luck I had in that regard.
1:55:40
And then just over time and it's been an interesting journey for me to grapple with my different position in in the ecosystem like so when I started the structury interviews that was sort of part of it which was I started out not knowing anyone.
1:55:59
I got to the point where I can talk to anyone that I want to and so how do I square that?
1:56:04
I can't be the guy with the chip on his shoulder trying to make a name for himself forever.
1:56:08
It sort of gets it's like the the meme with the guy how are you doing kids?
1:56:12
Like at some point you have to accept your part of the establishment.
1:56:16
How can I do that while still staying true to the idea that checkery is about the readers. It's reader funded. My loyalty is to them.
1:56:26
I'm very clear I have no loyalties to anybody else.
1:56:28
And so well I'll just I will talk to people sort of acknowledgement of what I can do but it's going to be fully transcribed and published and and sort of available to to everyone.
1:56:38
to everyone. Have you ever dealt with or thought about the attack vector of a special interest, you know, buying a thousand plus, you know, thousands of seats to a single, you know, independent publication and saying like, yeah, like,
1:56:52
you know, we're happy, you know, we we we got seats for all of our employees actually because we really, you know, love the and then and then suddenly they're sitting over there, you know, is representing meaning very meaningful amount of your revenue. 50%. 50%.
1:57:05
>> Um I mean uh I fortunately uh I think of of a scale that I don't have that problem. >> Yeah, it's good. >> There we go.
1:57:12
>> But yeah, it's uh but no, I think I think audience capture for subscription sites is a potential issue for sure.
1:57:18
>> And this is another thing I was sort of right place right time.
1:57:20
I got big enough by the time that it doesn't matter.
1:57:21
And >> if someone's really ups like I give refunds all the time actually if someone really upsets me, I will refund them and every dollar they paid me, I'm just like go away.
1:57:31
I don't you know I I I don't you're being abusive or whatever it might be.
1:57:35
>> And that that is a beautiful thing about the relatively low price high customer base model is no one has power over me.
1:57:45
like I I have the burden of publishing, you know, so as as often as I do, I feel a heavy weight of duty to my customers.
1:57:54
When I write something I'm not happy with, like I don't sleep well.
1:57:55
But at the same time, there's no one customer or no no individual that can come in and be mad at me and and >> yeah, >> impact my business.
1:58:04
Um I I I'm I'm seeing that there's maybe some sort of parallel between uh legacy media and independent media where uh independent media it it's not by default more pro tech or anything but there's just no salary cap.
1:58:18
So if you're at a legacy institution and you're writing probably some sort of rough loose salary cap of a few hundred,000 whereas you go independent it's feast or famine.
1:58:27
You might fail but you might get really really successful and have a huge income from that.
1:58:33
and and I'm wondering uh what we're seeing in the AI salary wars where we're seeing more and more talent and that you know Mark Zuckerberg potentially paying $100 million uh bonuses.
1:58:44
Um do you think that Apple will come around to spending more uh money on researchers?
1:58:51
It feels like they kind of have an internal salary cap with uh Tim Cook making 75 million.
1:58:55
There's now people that report two levels down from Mark Zuckerberg that are making more than Tim Cook.
1:59:01
And you have this weird dynamic where even if there's no actual salary cap at Apple, you kind of have an implicit one from the CEO. >> Yeah, for sure.
1:59:10
I mean, well, I think just to go back to to to the media observation you started out with, is as you increase transparency in the market, as you decrease non-related barriers, which in the publishing world previously was really geography.
1:59:25
And when everyone's on the internet, you inevitably in just about all cases, you get a power law distribution.
1:59:33
And a few people make a ton of money because they win most of the market.
1:59:38
And then some people make some and then there's a long tail that that sort of don't make any at all.
1:59:42
But it's it's very it it's interesting.
1:59:44
It's it's fluid in a way, but it can sort of become somewhat static as long as the people at the top sort of, you know, continue to do well.
1:59:54
But what's interesting about AI is for 40 years, you would have periods of time where you'd have tech companies going to head headto-head in a product market.
2:00:07
>> And I I think one of the reasons part of the software eating the world sort of idea is the way you get an apex predator is that that predator killed everyone else first.
2:00:17
And so you had tech companies fighting each other for the first 20, 30 years of tech.
2:00:20
The ones that emerged were lean mean killing machines and they and the entire industry were sort of set loose on the rest of the world and everyone was just like was is getting slaughtered sort of left and right.
2:00:32
But what you also had over this past sort of 20 years or so is the big companies in particular sort of slotting into unique slots.
2:00:41
So you have you have Facebook is is social, Google is search, Apple is devices, Microsoft is is business or you know business applications, Amazon e-commerce etc. Right.
2:00:54
Obviously, these companies are are very large and do lots of things and there's some overlap in different places, but they've been fairly sort of distinct in their categories and they've been dominant in those categories.
2:01:05
And so, they've been in a place where like Hollywood is wanting to get to, right?
2:01:11
What is the dream in Hollywood?
2:01:11
You want to have a franchise where the next Marvel movie matters more than who the star is.
2:01:19
The reason that's so great is because you now have bargaining power over the stars.
2:01:23
So you just sub someone else in.
2:01:25
And and whereas the old style like Tom Cruz makes the most money because Tom Cruz on a movie poster sells the poster.
2:01:31
And so in a negotiation, he has massive bargaining power.
2:01:34
So he's going to get get paid a lot get paid a lot of money. A in tech.
2:01:37
It hasn't been that case.
2:01:40
The companies themselves have been franchises.
2:01:42
And so the the overall anyone who works in tech or probably works in any any any entity, but you know, there's a few people in each company that are critically important, really make the whole thing go.
2:01:54
Everyone else is fairly replaceable.
2:01:56
Those people are have probably always been somewhat underpaid um for years and years and years, both just by the nature of companies and the cultural issues and your salary cap sort of analogy, but then also just like it's not a transparent market.
2:02:11
It's not it's not hard to price sort of what people are worth with AI.
2:02:15
Everyone's trying to do the exact same thing.
2:02:18
So you have multiple companies trying to do the same thing.
2:02:22
The output is somewhat measurable.
2:02:24
I mean all the AI test stuff has issues but by and large everyone kind of knows who has the good models and and who doesn't.
2:02:31
They you know the scalability questions you know like because all these companies are trying to do the same thing.
2:02:38
We have a very unique situation where the bargaining power, you increase transparency, you increase sort of the liquidity or the ability of people to move around because they're doing the same thing.
2:02:49
The bargaining power shifts to the people that are super valuable cuz suddenly it's much more clear who's valuable and their skills are much more transferable.
2:03:00
So this is I think a very underrated bare case for tech in terms of AI at least for this time period is they've lost that that murky bargaining power over employees that they enjoyed for decades >> and currently you're seeing what happens
2:03:20
when you don't have that you start paying employees what they're worth and obviously that's great I I'm not saying this this is a business analyst it's not a sort of a moral statement But it is like what Mark Zuckerberg is doing I think is totally rational. I think it's
2:03:32
I think it's a classic sort of Clayton Christensen from Facebook's perspective. AI is all upside.
2:03:39
So of course they're going to invest what they need to do to win.
2:03:41
But it's costing him a lot of money and by extension it's costing everyone else in the ecosystem a lot of money.
2:03:47
Well, isn't it in some way is the right way to think about the last couple weeks uh like more of like an aqua like an unofficial aqua hire in the sense that you're it's it's not just the the people, but it is the the knowhow in terms of hey here's there's these things that we want to do that are important to our business in a lot of different ways.
2:04:07
And we're basically it it's it's like the collective is actually more valuable than any one like the collective together getting 10 researchers at the same time is meaning you know is meaningfully more valuable than than than than just each individual researcher added up.
2:04:21
You know >> there's probably something to that but I I I think again like what is actually different between what Google is trying to do what Anthropic is trying to do what OpenAI is trying to do and what Meta is trying to do.
2:04:33
They're all trying to do the same thing.
2:04:35
So I my suspicion I'm not an AI researcher so I don't want to overstate my my knowledge in this space but my suspicion is skills are are fairly highly transferable and when that is the case there is in some situations if lots of people can do those skills that's terrible for the employees because then their bargaining power gets diminished because anyone can slot in.
2:04:59
But we're in this space where the skills are transparent, knowable, transferable, and there's not very many people that can do them.
2:05:05
And so it's it's a scarce resource that everyone's fighting over.
2:05:09
And that's why you see this real shift in negotiating leverage as as manifested through these dollar figures to to AI researchers. >> Yeah.
2:05:18
Do you think um I mean, Google seems like the most fragile and the most like paranoid about disruption. It's not all upside.
2:05:26
Uh it could be very bad for them.
2:05:28
um the innovator's dilemma.
2:05:28
You know, you had this back and forth where uh Cinder Pachai mentioned that he hadn't read the book.
2:05:35
You said it doesn't matter because it's a structural issue.
2:05:37
I think that's a good point.
2:05:37
But if you play back the counterfactual, is it ever possible to disrupt yourself and essentially like if the Gemini app had launched before chat GPT and they had taken over that mind share and maintain 90% ownership in that like it would be somewhat disruptive to their revenue and their profits as they transition over.
2:06:02
But when I sum the revenues from OpenAI and LLMs and then Google search, I'm not seeing some massive drop off that's actually that actually would destroy Google in the medium short to medium term.
2:06:16
So, but I'm wondering if you think it's like is it entirely impossible to avoid the innovator's dilemma by disrupting yourself?
2:06:24
>> Well, number one, you have to also look at margins, not just revenue. Yeah.
2:06:26
Um, but number two, >> you actually you answered your question.
2:06:32
Google didn't launch Gemini as a chatbot. That's the answer. They were years ahead.
2:06:36
They they invented the transformer a decade ago.
2:06:41
>> And and so in many respects like there's parts of this question that the counterfactual makes the point in that it is a counterfactual and it's not reality.
2:06:53
>> Now, I do think I think Google's done better than I expected over the last two years.
2:06:57
Uh I I like what they're doing in search generally.
2:07:01
I I think they it does seem to be the one part of the company that still functions like they they can actually iterate and build products.
2:07:10
What we're seeing is reminiscent of what they did a decade 15 or 12 years ago when everyone's like vertical search Google's done all the everyone's going to search in apps and Google completely transformed the SER the the search uh engine uh response page uh whatever it is uh the search engine results page to
2:07:26
be local or to be shopping or whatever and Yelp's been throwing a hissy fit sort of ever since and and so that's what they're doing with search right and and with search overviews and they have this new search labs or or AI I mode they can sort of test stuff out once it's scalable. Once they they're
2:07:42
Once they they're confident about the monetization issues, they can sort of shift it over.
2:07:44
I called it the search funnel, search AI funnel.
2:07:48
I think it makes a lot of sense.
2:07:51
And I think and I this has always actually kind of puzzled me where I think they're responding fairly well even though this is seems to be a textbook case of disruption.
2:08:01
And I went back to an article I wrote years ago uh called Microsoft's Monopoly Hangover.
2:08:07
And I was I I I went through Lou Gersonner's autobiography and about how he turned around IBM and his real insight with IBM was everyone wanted him to break it up in into sort of different pieces.
2:08:22
And what he realized was IBM was so big and and large from having downstream a bit monopoly that actually the only thing they were good at was being big.
2:08:32
And so breaking them up would actually just create a bunch of subscale low-performing companies that would all get wiped out.
2:08:40
But as this behemoth, they could go to other big companies and solve all their problems at a very mediocre level.
2:08:48
But it still is sort of an attractive proposition.
2:08:50
And under Gersonner, they really rode the internet wave.
2:08:54
They went to all these big companies said, "This internet thing's happening. You need help.
2:08:57
We'll solve your problems for you."
2:08:59
and had a very sort of successful run, you know, kind of until cloud came along and which Gersner, by the way, was was was a proponent of, but you know, by that time the IBM people were back in charge and I was thinking about the the context of Microsoft where m you business models are hard to change and disruption is ultimately about business models >> and culture is hard to ch even harder to change.
2:09:25
But what can't really be changed is the nature of who you are.
2:09:28
And a and I think there's you know in Microsoft they were in a similar situation.
2:09:33
they were in a similar situation. They were a big monopoly and they weren't a product company and the attempts to become a product company with Windows 8 and all the things that went on around that time inevitably inevitably failed
2:09:46
and Sai Adella to his great credit and you know sort of diminished Windows importance in the company broke it literally broke it into pieces spread it around and this was a multi-step process and and got Microsoft back to a place of we're big and we'll do everything. We're
2:10:03
We're we're we're not a Windows company.
2:10:06
We'll go in there and we'll go solve all your problems.
2:10:08
Very sort of reminiscent of of the the second version of IBM.
2:10:10
And I go back to Google and I've always been intrigued by the I'm feeling lucky button, which doesn't exist anymore, but I always enjoyed that that button continued to exist long after you it was impossible to click >> because the moment you started typing the search box, it would start auto searching immediately and jump jump right to a search page.
2:10:34
right to a search page. But it was it was there in a it's just so core to Google to give you the answer to to know everything like the to to know everything about the world and to there's a bit where even though the core
2:10:51
of their business model is 10 blue links and it's not just the the users choosing the search link which gives them a data feedback loop so they know which results better but also the users choose the winner of an auction Google puts on for ads and it's an incredible business model. And there's something about that
2:11:05
And there's something about that that's always been intention and counter to what Google was founded to be.
2:11:10
And I feel like that germ of what Google was founded and meant to be is an AI answer engine.
2:11:20
And and it almost feels like even though Google is old and large and fat and slowm moving, that core aspect of their nature and is is still in the culture and that's why they're finding it in themselves, I think, to do better in AI than you would expect.
2:11:40
Was it enough to launch a chat GPT before open AI? No.
2:11:45
>> Uh was it uh was it enough to have any sort of cogent response for the first 6 to9 months? No.
2:11:49
to9 months? No. But it was enough that I think they've done better than I expected over the past year in particular and gives me I think more optimism than I expected I would have for the company when you know I when chat GPT first launched
2:12:09
>> AI overview from Google if you search Google's mission Google's mission is to organize the world's information and make it universally accessible and useful which is exactly what language models do really really well like the thing that's just undeniable, right? It's you can you can debate whether uh
2:12:21
It's you can you can debate whether uh this is going to be the year of agents.
2:12:26
It doesn't feel that way to me yet, but this is the year that >> most people have realized that wow >> LLMs are very good at organizing surfacing and and making data valuable. >> Yeah.
2:12:40
You you mentioned uh just the the debate over breaking up IBM.
2:12:44
Uh I'm interested if you could take us through >> I bet you didn't realize you're going to be talking about IBM today, did you? >> No. No. No.
2:12:50
Uh, but I want to I want to talk about Intel and and kind of your the the history of some of your takeaways and what you think you've gotten right in the past, your perception of, you know, should they break up the the foundry business uh and what you think might be in the works with Lip Bhutan coming in there.
2:13:07
Um because it I was listening to Dylan Patel talk about his conversation with uh the new CEO Lip Bhutan and it seems like they're doing lots of tightening up, lots of layoffs, but uh it's kind of I I don't even know what framework to apply to analyze like is a breakup the correct thing.
2:13:27
It feels like something people just say. >> Yeah. Um so Intel, it's funny.
2:13:30
>> Yeah. Um so Intel, it's funny. I one of my very first articles was about Intel >> and what I said at the time was and this was 2013 and this was an art like you know when you start a site like you're like a new band and why does everyone think a new band's first album is the best cuz they've been working on these
2:13:53
songs for years right and then the next album they had a year to do it and they all suck right so I'll let people decide if that applies to or not um I won't be offended more slow >> but Yeah, but I'd have been on, you know, Intel had been a thing I've been wondering about for a long time, which was by 2013 when I started, they had clearly missed mobile. Now, it wasn't
2:14:11
Now, it wasn't clear to them.
2:14:13
They were still trying to do the Atom processor and and just they're going to figure it out tomorrow.
2:14:20
And the the problem with missing mobile is the problem with Intel in general is Intel is always very biased towards high performance.
2:14:29
And this goes back to uh actually Pat Pat Galsinger his first time through at Intel.
2:14:36
>> Intel, you know, had the CISK uh uh the the way there's CISK versus risk.
2:14:40
It's like uh it's different ways of organizing bits or whatever, risk is generally more efficient.
2:14:46
And actually even Intel processors today, even though x86 is CISK, the internal it's re-ransated internally to a risk type language.
2:14:55
language. Um, none of that is really important other than to say in the 80s there was a real push in Intel to switch away from x86 and to to a risk type um uh of uh I don't use but like um for for the processors >> and Gaussinger was a leading proponent
2:15:12
that this is a terrible idea >> and the reason it's a terrible idea is because there was already a huge ecosystem of software built around x86 and all this low-level code and capabilities that no one ever that was written once and no one ever wants to touch again because it's miserable work. And he's like to rewrite all that stuff
2:15:30
And he's like to rewrite all that stuff would take at least two years.
2:15:37
And in that time, our ability to manufacture chips will improve so much that had we just stuck with CISK, our processors would be faster. >> Mhm.
2:15:48
>> And that was the right bet.
2:15:48
And that's one of those foundational bets that I why I like to think about companies and their history and what goes into that which is Intel from the 80s on has solved its problems by having superior manufacturing and by moving faster.
2:16:01
And yeah, our chips may be theoretically less efficient, but if our manufacturing is better and our transistors are smaller, it doesn't matter cuz that will swamp whatever theoretical sort of efficiency you might have.
2:16:13
And this drove the entire computer industry.
2:16:15
you you you you would write to write a program to every second you spent optimizing your software in the 80s or '9s was a waste of time because whatever improvements you could get would be swamped by the next version of uh if you went from 286 to 386 or 36 to 486.
2:16:31
That jump was so large you were better off focusing on features even if it made your software sort of slow to use on the current hardware because the next generation of hardware would be so much faster.
2:16:44
it would solve your your speed problems for you.
2:16:46
Now, this has generated a lot of bad habits amongst tech developers.
2:16:50
That's why you get bloat and why you have like poor performing things and all those sort of things.
2:16:55
But that this was sort of super critical.
2:16:57
And so, Intel at its core has always been focused, they've always been manufacturing first and focused on better and better performance.
2:17:02
What happened with mobile is in that calculation did not come efficiency.
2:17:09
They were never focused on efficiency and in mobile efficiency was everything.
2:17:14
So what happened with mobile is Apple Apple went with an ARM processor made by made by made by Samsung and they basically rewrote everything.
2:17:21
All that stuff Intel didn't want to rewrite in the 80s or if they rewrote would just give other process processor companies a chance to catch up with them had to be rewritten for mobile because efficiency was so much more important than performance.
2:17:36
When that happened, Intel was screwed.
2:17:38
Now, it took them a long, long time to realize they were screwed, but they they were just fundamentally unsuited to be competitive.
2:17:43
It was the whole Paul Adelini turning down the iPhone contract is not true.
2:17:51
Tony Fidel, I I I said that once and I got a call from Tony Fidel actually that this when I had him on it for an interview and he's like, "This drives me up the wall.
2:17:58
Intel was not remotely competitive even though they had ARM chips then.
2:18:02
Even their ARM chips then were focused on performance, not on efficiency."
2:18:05
And and so the the problem for the problem for Intel is once you missed mobile, you were going to lose your manufacturing lead at some point because volume matters so much.
2:18:16
And every time you move down the curve, your transistors get smaller.
2:18:21
The costs increase massively.
2:18:23
So you need volume to spread out the cost of building these fabs.
2:18:28
Like back then when I wrote this article, fabs cost 500 million.
2:18:30
Now they cost like 20 billion.
2:18:32
And this is over a course of like 12 years. Mhm.
2:18:36
>> So, so it was clear Intel was going to be in big trouble back then.
2:18:39
And so I wrote they need to build a foundry business.
2:18:43
They need to figure out a way to build chips for other people because in the long run the cost of keeping up in manufacturing is not going to be tenable if you're not making mobile chips.
2:18:55
And what obviously they didn't.
2:18:57
TSMC made all the mobile chips for everyone. And guess what happened?
2:19:00
TSMC took over the manufacturing lead.
2:19:02
Now, there's lots of other things that went into this, why Intel stumbled and sort of things, but at a structural level, what happened was actually inevitable once Intel missed mobile, unless they figured out a way to make mobile chips some other way. They didn't do that.
2:19:21
>> What's interesting is >> what is the problem with that, it took so long to manifest.
2:19:30
Part of mobile was you had an explosion in the cloud because cloud and mobile actually go hand in hand.
2:19:33
Intel made all those cloud ships.
2:19:35
Intel stock had an incredible run from the time I wrote that article for the next 8 to nine years.
2:19:40
And I felt like kind of a cuz I'm saying this company is screwed if they don't do what I say.
2:19:43
They didn't do what I say and their stock went to the moon.
2:19:46
But what the the way it actually caught up to them has been in the past two to three years where there's astronomical demand for AI chips. Only TSMC can meet it. Intel's not in the game.
2:19:59
They're they're trying to shift to a foundry model, but they're they're so far behind it.
2:20:03
Being a foundry is being a customer service business.
2:20:07
It's not being an Intel we tell you what to do or we we tell our design teams how to change their chips to accommodate our manufacturing needs.
2:20:16
It's just it's totally different.
2:20:16
And they needed a decade to learn how to do that.
2:20:20
Had they changed in 2013, they would be ready today to capitalize on AI.
2:20:27
And and the counter example here is Microsoft.
2:20:31
Microsoft building Azure.
2:20:31
Yes, it got them somewhat in the game with mobile and things like that, but AWS dominates uh in that space, but by virtue of building up Azure, they were prepared when the AI opportunity came along.
2:20:46
And now Azure is is sort of a big AI player.
2:20:51
And you know, I wrote about these these two examples a few weeks ago in the context of Apple.
2:20:55
I think the concern for Apple isn't the short term.
2:20:57
We're going to be using AI apps on our iPhones for quite a while.
2:21:04
It's are they going to be prepared for what's next if they don't do some sort of sort of reset and pivot here?
2:21:13
>> Oh, sorry to answer your question about Intel. any.
2:21:16
>> Yeah, I mean it's like a managed decline basically like just like you know just get as much cash flow out of this thing as you can while you wind down the business >> for Intel.
2:21:26
>> Yeah, that's what I'm hearing.
2:21:26
It doesn't feel like oh yeah there's a silver bullet just split the business and they're good.
2:21:31
Like no it's like it's it's >> the reason not to split the business is Intel needs volume and they get volume from Intel.
2:21:36
uh and uh the and AMD split their business a decade ago and it was >> really they had a very hard time for many years and they had very tense and difficult negotiations between the global foundry side and the AMD side.
2:21:51
Global Foundaries was AMD's uh uh manufacturing arm.
2:21:53
Um and it wasn't until really they got out of that and went to TSMC and then also completely rehauled their ship design business uh and all those you know um that they they got in the business they were and then also that Intel stubbled uh that that certainly really helped them Intel today.
2:22:08
So if you split it up like who's buy like Intel's Intel itself is fabbing some of its stuff with TSMC. >> Yeah.
2:22:17
>> Who who wants to buy Intel's foundry services?
2:22:19
The the problem here is uh TSMC is located in a country called Taiwan.
2:22:23
Um which >> you know what it is today, but 5 years ago it' have been like what Thailand?
2:22:27
Um which by the way was probably much better for Taiwan security when the American thought it was Thailand.
2:22:32
Um but >> so there's a real national security element here and it's just it's a really tough situation because Intel is a failed company at this point and they're and the the reason the failure is so total is because the aspects that drive their failure are the same things that drove their success. It was their arrogance.
2:22:58
It was their a sense that we're the best, that we will just win through manufacturing might and performance.
2:23:07
And all those things work against becoming a good foundry, work against being a customer service organization, work against recognizing the fact that you're not going to make up for missing mobile through manufacturing, which was their bet for years and years.
2:23:21
You you had to accept that you lost.
2:23:23
And and that's a tough place for companies.
2:23:27
It's not like someone made a mistake.
2:23:29
It's that what they did what they did too well for too long >> is who they were.
2:23:34
They continued being who they were. Right. That's right.
2:23:36
But who else are you going to get if you want an alternative to TSMC?
2:23:40
It's it's it's a very situation.
2:23:44
>> Last question and I think we'll be forced to to have you make a slightly shorter answer.
2:23:48
Unfortunately, I wish I wish we had hours to keep talking.
2:23:50
I wanted to get your updated thinking on XAIX, the combined entity.
2:23:53
The last 24 hours have been very chaotic.
2:23:57
When the initial merger was announced, it made sense for financial reasons for some of the different stakeholders, but I wasn't fully sold on this idea.
2:24:06
Uh, how >> you're going to force me to come with takes that I I I generally just avoid right about Elon Musk companies um for uh self-sanity reasons.
2:24:16
I think I mean I remember I wrote an article years ago about like uh when the Model Y was announced >> and I was talking about you know it's a Tesla and this aspect.
2:24:25
What Elon Musk is very incredible at is sort of creating reality out of thin air.
2:24:31
Um he's like the ultimate memer and to create like um like it's the way things used to work backwards.
2:24:43
Uh I remember I analogized it to like protests like a critique of of of modern protests is they spin up very quickly because social media makes it very possible but there's no infrastructure under them so they don't amount to anything.
2:24:53
Whereas you go back to like the civil rights era there was years of groundwork that went into like the million man march uh you know on Washington DC and there was a structure in place that ultimately manifested in large crowds.
2:25:06
But modern protests are the opposite.
2:25:08
The largess comes at the beginning and then it all falls apart.
2:25:11
there's nothing in place and um >> there there's something the that makes it challenge to write about anything Elon Musk related is the you have all the social aspects is you have this bit about Tesla of creating reality it's the stock was butressed for years by these
2:25:29
true believers even though the financial parts didn't make sense you famously had these wars with the short sellers and all that sort of thing and it worked it basically manifested a market for this Model Y and then And then the Model X, uh, not the Model X, what's the other one? Um, Um, >> the Model 3. Yeah.
2:25:44
So, it was Model 3, sorry, when I wrote that article.
2:25:46
Model 3 and Model Y had this massively successful and all all the people that were true believers got very rich.
2:25:52
Uh, and congratul congratulations to them. I It's great.
2:25:58
>> But it makes it almost impossible for someone for what I do who I want to look at structure and fundamentals.
2:26:02
I can observe this effect happening, but you can't really say what's going to happen or the effects of it other than to say this is interesting.
2:26:10
And so I wrote about that article and then the Solar City thing came out and he's like bailing out like his brother-in-law or something and I'm like I can't write about this.
2:26:19
Like what am I going to say?
2:26:21
Like like there's it just doesn't make sense.
2:26:22
And so I think there's to fast forward to X XAI.
2:26:27
Um yeah, there's a theoretical piece here.
2:26:30
I think actually XAI would be an incredible acquisition target for a lot of companies if it wasn't saddled with X. Uh so interesting.
2:26:38
>> It feels like the end state is e like Twitter getting spun out again.
2:26:41
Like that that that's my I that's kind of like my my it just ends up going back to Twitter and and and it becomes >> the blueber.
2:26:51
No one actually wants to like Twitter Twitter.
2:26:53
There's never been a company in the history of the world probably where the impact of a company is completely and utterly divorced from its financial realities.
2:27:02
Like I think when Elon Musk bought it and I assume that's continued through now, they'd had like one profitable quarter in their history.
2:27:11
Like it it's an unbelievably terrible business.
2:27:13
Uh and so I think it's probably weighing XAI down.
2:27:16
There's a yes I get the theory that Twitter data helps XAI contract that data.
2:27:22
You don't need to pay 43 million for Twitter to to to or 43 billion I should say to to get it.
2:27:26
So >> yeah, that that was always my position too.
2:27:30
I don't think it helped yesterday when when Mecca Hitler emerged on the timeline >> on the timeline, but uh >> good luck.
2:27:37
Hopefully they sort it out.
2:27:38
>> I wish I wish we had uh a lot more time here, but hopefully we can do it again.
2:27:42
>> Thank you so much for stopping by. >> Yeah, no worries.
2:27:44
Uh I love what you guys are doing.
2:27:45
I actually had the idea of doing a daily podcast ages ago.
2:27:47
Um, classic example of ideas don't count, execution does, and you guys you guys did it. I think it's great.
2:27:54
>> Well, you're always welcome here.
2:27:55
>> You're always welcome. Thanks so much. Thank you.
2:27:57
>> We'll talk to you soon, man. >> Bye.
2:27:59
>> Meta just uh is going deeper with Rayban maker uh eslxotica.
2:28:05
I I I cannot pronounce that first word, but people just call it.
2:28:07
Um, and so Meta is taking a minority stake uh in Lxotica to accelerate its smart glasses ambitions, investing $3.
2:28:15
5 billion in the iconic Ray-B band manufacturer.
2:28:18
Uh, we were talking to David Center about the the history of this company. It is fascinating.
2:28:24
I'm very excited for him to uh break it down for us a little bit more.
2:28:28
Hopefully he can come on the show and and talk about it because it's >> very very soon.
2:28:31
The founder has a crazy story.
2:28:33
story. He grew up in I think he grew up in an orphanage >> and uh and and just what do they wasn't they didn't call him the pit bull they called him something else but he was an absolute savage apparently at one point he wanted to buy Oakley and the um and
2:28:51
the the founder CEO of Oakley didn't want to sell and so the CEO of Luxodica acquired a the largest retailer for Oakleys and just pulled them off the shelf and basically and started selling knockoff Oakley even though they were trademarked. And then eventually the
2:29:04
And then eventually the Oakley CEO came around and said, "Okay, like I'll sell.
2:29:08
You're cratering, you know, my revenue. Let's do a deal."
2:29:12
>> Um, so absolute dog and uh soon we'll have to break it down.
2:29:18
>> What do you make of this idea that like, you know, Apple when they make a device, they they they redefine and very much standardize that particular market.
2:29:29
So when they come out with watches, there are a number of styles of watch.
2:29:33
There's the dress watch, the sports watch, the steel sports watch.
2:29:37
There's the dive watch, there's the, you know, Casio style.
2:29:41
There's a whole bunch of different styles, right?
2:29:44
>> Apple comes in and just says there's only one style, the Apple Watch, and they become the number one Apple style.
2:29:49
>> And they give you some variance in the band, >> in the band, little stuff here and there.
2:29:53
And they were doing partnerships.
2:29:55
I think they did Hermes band for a while.
2:29:56
They've done a couple other things, but it's been mostly Apple's design language on your wrist.
2:30:03
>> Whereas with the Meta Ray-B bands, they're saying, and now the Meta Oakleys, they're saying you like the look of Ray-B bands.
2:30:09
We're just putting our technology into the style you like.
2:30:13
We're not going to try and create a new iconic style that says meta like Apple says headphones.
2:30:19
says headphones. um and and they're just kind of like they're very very different strategies and and so it feels like >> well so so I think this is strategic this doesn't mean that uh this doesn't mean that Meta can't develop their own styles in time but I think it's very smart to say
2:30:36
>> hey we don't need to innovate on aesthetics and the sort of silhouettes right there's classic silhouettes Ray-B band silhouette is Lindy these Oakley silhouettes are very lindy >> and they're different markets the market lxodica has I think Garrett late and like a bunch of other like um brands under it. So they're basically saying
2:30:53
under it. So they're basically saying like through this we can deliver Luxodica has >> brands in every for every demo that you that Meta could possibly want right as a as a hundred billion dollar >> you know company >> and so I think it's very smart I think uh uh Apple like you said will will
2:31:11
probably take a a drastically different approach in terms of like standardizing around something and and that will say something but accessories like eyewear are just such a such a personal decision and such an expression of of um of who somebody is that I think that uh you want to give people max amount of optionality. >> Yeah. It's just interesting because like >> Yeah.
2:31:28
It's just interesting because like you could have said that about watches like you before the Apple Watch you could have said that well you know somebody who wears a dress watch wants a dress watch.
2:31:37
Somebody who wants a steel sports watch.
2:31:39
Somebody who wants a G-Shock is G-Shock. It's like the G-Shock.
2:31:42
You say G-Shock and you just immediately think like you know special operations guy or Jaco Willink listener like that that that it's like a durable rugged thing.
2:31:51
you say, you know, Rolex, that's a different thing, right?
2:31:54
Uh and and Apple was able to standardize around it.
2:31:58
And it's interesting that that uh Meta hasn't been trying to do that and instead they're they're focusing on partnership here.
2:32:04
It's just like a it's just an uncommon strategy, but it seems to be working.
2:32:07
Um I there's another post in here.
2:32:10
I don't know if we have it here, but >> I'm trying to think of a new like the key the key thing is >> Apple's great at at innovating at multiple layers, but like gen generally it's very hard to try to deliver hits in like two specific areas like aesthetics and design >> and then simultaneously in something that's basically a fashion product and then simultaneously deliver the technology. >> So I don't know. >> Yeah.
2:32:35
Uh Jack Ray here says, "After wearing Rayban Meta Wayfairer glasses for a few weeks, I feel kind of naked wearing regular sunglasses.
2:32:41
I found three use cases that are hard to roll back.
2:32:45
One, spontaneous photos of my kids when we're out and about.
2:32:47
Any cool pose that has a halflife of 3 seconds I can now capture instead of pulling out your phone.
2:32:53
Uh optionality of music or hands-free phone calls without digging around for earbuds.
2:32:58
Uh and three, knowledge seeking chat when I'm walking around, usually for simple factual things.
2:33:04
That's exactly what I experienced when I was uh demoing the Rayban uh meta waveferrors.
2:33:08
Um it turns out there's more questions I feel like asking when there's no friction.
2:33:13
I'm very excited for multimodal and real-time translation use cases too.
2:33:18
They're only going to get better.
2:33:18
But I think those three are maybe enough.
2:33:19
And I I think with a lot of these products, just having one killer use case, like just replacing the the, you know, the headphones for hands-free phone calls or something like if you can just become someone's daily solution for music, like that's enough to just sell the product and then sell them another one the next year when it upgrades a little bit.
2:33:41
Sell them another one, keep them as an active user and and roll that out for a long time.
2:33:46
And then if they can do the other stuff, that's great, too.
2:33:49
really nail this stuff, but it's fascinating to see them.
2:33:55
And it's also interesting how it feels like Google was uh was talking about getting into this this space.
2:34:00
We saw some launches at IO.
2:34:02
Haven't actually seen any of those in the wild.
2:34:04
Haven't seen anyone really talking about those.
2:34:07
Uh Apple, it feels like this would be something that they could jump forward to with a stylish pair of eyeglasses with some basic functionality.
2:34:13
just take what's in the AirPods, take a camera, like they could do something cool.
2:34:18
Um, but they're like just much slower than than >> Yeah.
2:34:24
The other the other thing with eyewear that's different or that's going to be like a new challenge for manufacturers is that there's so many different situations where I might want to wear something like a Ray-B band >> or or a Jam silhouette >> one day and then I might want to I'm >> J Marie Mage. Okay.
2:34:41
Um but um the uh >> you know and then that same afternoon I'm wearing Oakleys when I'm playing tennis or something like that and and so there's a lot more like swapping and then then obviously >> I mean if they can keep the price low you could maybe wind up selling people multiple pairs and have indoor pair outdoor pair.
2:35:03
It's it's kind of inconvenient.
2:35:05
I feel like there's got to be a better solution to that but I don't know >> what's those what are the >> Yeah. the bif focals. >> Yeah.
2:35:12
Where they can flip down.
2:35:13
>> There's transition lens lenses, but those never fully work all the way, but then there's the flip down ones, clipons.
2:35:18
There's all sorts of different solutions.
2:35:20
The big news is that the third browser war has begun.
2:35:23
Um, Google stock has dropped on the news that OpenAI is planning to launch a a Google Chrome competitor within just weeks.
2:35:32
And this is very interesting timing because >> it's time to browse. >> Yeah, time to browse.
2:35:40
certainly makes sense to become deeper in more deeply integrated into the user's life. Makes a ton of sense.
2:35:46
There's a ton of benefits that come from having a web browser.
2:35:47
Um what was interesting is uh the we can go into what Google actually la or what OpenAI is talking about launching, but this news this scoop leaked the same day that uh Arvin from Perplexity announced that they're finally releasing their next big product after launching Perplexity.
2:36:08
Comet, the browser that's designed to be your thought partner and assistant for every aspect of your digital life, work in personal.
2:36:13
And so Perplexity launched this on June 9th.
2:36:15
And then OpenAI, the the scoop goes out via Reuters the same day.
2:36:21
And so this feels like very much like let's not let Perplexity get a bunch of attention and drive a bunch of people to to start daily driving comet the browser because even though we're not ready to launch our competitor, we want Arvin was on the show talking about Comet, but over a month ago, he said it was really important to the business.
2:36:43
This was a big bet that they're making.
2:36:45
Uh he uh and I'm sure both companies are racing to be the first to launch, but Dia the browser from the browser company uh also launched out of or they're still in beta, but they launched like a month ago or something like that.
2:36:57
So, >> this is you know, you're not going to be the first.
2:37:00
>> Oh, they launched a month ago with the DA browser.
2:37:01
interesting because I saw Riley Brown also posted the cursor for web browsering DIA browser and I thought Dia browser launched that same day but I guess it had launched earlier.
2:37:08
Um >> yeah so anybody that was an Arc user can download DIA today >> uh and chat with their tabs but interestingly enough Perplexity's brow browser and open browser are both built on Chromium the same open source project that underpins uh Google Chrome and Microsoft Edge. >> Yeah.
2:37:27
So it the cool thing here that means that they're compatible uh compatible with uh existing Chrome extensions. >> Oh, interesting. Interesting. Okay, that's cool.
2:37:37
Um yeah, it's it it's I I I want to talk to more people who were like active in tech during the earlier browser wars.
2:37:45
The first browser war was Netscape Navigator versus Microsoft Internet Explorer.
2:37:50
This is in the mid mid 90s, early 2000s.
2:37:52
Uh Netscape was super dominant and everyone loved Netscape.
2:37:57
It was originally the mosaic browser.
2:37:59
This is the Mark Andre project and then but Microsoft bundled Internet Explorer with Windows 95 and the distribution was so powerful that Internet Explorer actually wound up winning and became really really dominant.
2:38:11
But then there was this lawsuit and went back and forth but then uh basically in by the early 2000s Internet Explorer had over 90% market share but then they got kind of lazy and stagnant apparently.
2:38:23
And I mean I'm I'm not exactly sure ex what happened but they there was a lot more competition.
2:38:28
So Firefox which was I believe like a spinout of Netscape or kind of like some of the same heritage there.
2:38:33
Um began getting traction and then Google Chrome launched in 2008 and leaprogged everyone.
2:38:39
Uh and Google Chrome was really focused on like speed.
2:38:41
It was the fastest browser.
2:38:42
Um and they they did a whole bunch of work to optimize JavaScript so the pages would just load faster and run better on pretty much every computer that you had.
2:38:50
And so uh and then they had the open source project with Chromium.
2:38:53
And so they were able to kind of standardize the entire industry.
2:38:55
And so everyone's always been trying to draw, uh analogies between like the browser wars and the LLM wars and like what's the role of open-source in that?
2:39:04
Like is open source a strategy to wind up maintaining your your dominance?
2:39:10
How much does distribution matter?
2:39:11
like Chrome was probably pretty easy to distribute because every single person was visiting Google just every day searching and so you just put this bar hey want to switch to the faster browser and people just do it because you have basically like you know billions of ad impressions on your product every day.
2:39:29
>> Will be interesting to see if chat GPT can get people to download their own browser on desktop.
2:39:34
I mean, I'm using chat GPT on desktop in Chrome all the time.
2:39:39
>> Which chat GPT model would you want to use as a default search engine?
2:39:44
>> That's the hard part because I always run into this problem where it defaults to 03 Pro, but that takes 10 minutes.
2:39:51
And so then I have to go to 40.
2:39:51
And then if I'm in an 03 Pro flow and I'm talking to 03 Pro and I let it cook for 10 minutes, it gave me a great answer, but then I want to just be like, okay, just like clean this up a little bit or summarize this or do some bullet points. I want 40 to do that.
2:40:04
So, I have to switch over. So, I don't know.
2:40:06
I I would imagine I'd go 40 as the default because I want speed.
2:40:10
But even 40 could probably be faster before it truly replaced. >> Google's very fast.
2:40:16
They've spent a very long time being fast. >> Yeah.
2:40:20
And I could imagine them doing a similar project to I believe it was like the V8 JavaScript engine.
2:40:24
They sent this team out to uh I want to say like Iceland or something.
2:40:30
Uh they they basically sent like a bunch of engineers to like an offsite and they were like just go optimize JavaScript for like a month.
2:40:41
Just go focus on this for like a month or months and come back when it's done.
2:40:45
Like you have no other responsibilities than just like optimizing this like compiler.
2:40:47
And they came out came back with the V8 JavaScript engine.
2:40:50
It created this whole like Node. js boom.
2:40:52
People were running JavaScript on the server then.
2:40:53
And uh and I could see Google kind of doing something similar where they're like, "Okay, we have Gemini.
2:40:59
It's good at looking stuff up.
2:41:02
It's a good knowledge retrieval engine.
2:41:04
Go figure out how to make it load all the tokens for the full response in 100 milliseconds.
2:41:09
And that would be very very cool.
2:41:13
And I wonder if that's like a uniquely Google advantage.
2:41:16
Tyler, you look something up.
2:41:18
>> Yeah, it was in it was in Denmark. >> Denmark. Okay. I was close. I was close.
2:41:21
Yeah, I wasn't sure it was Finland or Iceland and Denmark. >> Yeah.
2:41:24
The interesting thing here, I'm realizing that tabs are definitely a light lock in to browser.
2:41:29
It's not just the default, but if you have six to 10 tabs that you've just had for a really long time and they're like a bunch of different things and you can't exactly remember what they were if you had to list them all off, but you know, you know, I I personally end up using tabs as like >> somewhat of a to-do list.
2:41:48
>> And so if you're spinning up a new browser and you don't have your tabs, it's like, oh, do I want to just like get rid of my my tab stack?
2:41:54
I have a bunch of tabs that just have stayed there for years and they're basically like it's basically like a mini operating system, right?
2:42:02
With like different apps that might be >> a Google sheet or something else. >> Yeah.
2:42:07
No, I know what you mean.
2:42:08
>> So, there's very real lock in.
2:42:08
I could bring all those tabs over, but I have to then >> log in to a bunch of different services.
2:42:16
And so, it's it's really really hard to actually >> Yeah. >> uh win here.
2:42:19
I wonder if anyone's using, you know, in in Google Chrome, you can actually change the default search bar to, you know, when you type in the search bar and if you just type words, it just Google searches it.
2:42:28
You can change that to search chatgpt.
2:42:33
>> Yeah, like you can pass in a query parameter and it can just do that.
2:42:35
But I haven't heard of anyone actually doing that.
2:42:37
And I used to have I used to be such a power user of Chrome.
2:42:39
I used to have different code words basically.
2:42:41
So if I if I typed like I space and then a query, it would go to IMDb and search that specifically.
2:42:51
So you could you could have Chrome like route to any specific search any you could press like Y space and it would search Yelp or you know anything else.
2:43:00
Um but I don't know if people are I don't know if people are doing that with Google with Chad GPT.
2:43:04
I think people mostly just like control command T and then hang out in Chad GPT.
2:43:09
Well, we'll have to ask uh Chris in 15 minutes about get an update on the browser wars because he was uh an early investor in >> I know one of those tabs that you have pinned right now. >> What's that? >> Adio, >> of course.
2:43:22
>> Customer relationship magic.
2:43:22
Adio is the AI native CRM that builds, scales, and grows your company to the next level.
2:43:27
You can get started for free.
2:43:28
>> I've had Adio open for thousands of hours in a row at this point. >> Yeah.
2:43:34
Uh so, Signal kind of breaks it down with the Open AI launching the web browser.
2:43:38
says, "This is the oldest play in TA.
2:43:39
Find product market fit with a single killer use case.
2:43:41
Then vertically integrate and horizontally expand until you control the interface layer itself. App platform.
2:43:46
Once you own the interface, you own the defaults.
2:43:48
Welcome to the next generation of browser wars." I Yeah.
2:43:52
I Yeah. What's interesting is there like Sam Alman at OpenAI and just the fact that OpenAI is a company like there is kind of a mandate to like vertically and horizontally integrate figure out code figure out research figure out devices but every company wants to do everything
2:44:11
but then sometimes they run up against barriers like there was a time when Google was like we want to win social networking and we want to beat Facebook and we're going to launch a direct Facebook competitor and they did and it didn't go well and
2:44:25
then they shelved it and then they wound up producing trillions of dollars in market cap just doing the thing that they do great and so the question is like the surface area of open AI they have to exp explore they have to experiment it's it would be stupid not to see if they could get a browser and a
2:44:41
device and a chip and a nuclear reactor and everything and sand get the get the sand get everything um but but there's no there's no guarantee that they will win the entire vertical stack and will be the one company, right? >> I think my question is are these going
2:44:54
>> I think my question is are these going to be like is OpenAI's browser going to be an entirely new app other than their existing >> mobile app?
2:45:02
Is it is or their desktop app?
2:45:05
app? I yeah that is interesting >> because if they have to get people to reddownload a separate app then then that's then that's like an entirely you know they have a good fly you know they have >> they wouldn't just evolve the apps they already have
2:45:18
>> perplexity too I don't I don't perplexity has uh is planning to to release this as like a new standalone app or it will be in the perplexity >> mobile app but >> yeah um yeah I mean I know I think comments like its own thing cuz we were looking to download it and we need a code. Um,
2:45:35
Um, and you can't just get it if you're just on perplexity. >> Um, but I don't know.
2:45:42
>> All I know is that you should go to fin.
2:45:43
AI, the number one AI agent for customer service, number one in performance benchmarks, number one in per competitive bake offs, number one ranking on G2.
2:45:51
>> Um, so, uh, Arvin breaks down like his philosophy of, uh, of Comet, the browser that he's dropping from Perplexi.
2:45:57
He says uh you can either keep waiting for connectors and MCP servers for bringing in context from third party apps or you can just download and use comet and let the agent take care of browsing your tabs and pulling relevant info.
2:46:09
It's a much cleaner way to make agents work. So that is interesting.
2:46:14
So I wonder how much like puppeteering will be in this because Chachi Chacht and OpenAI have operator that operates a Chromium front like a headless web browser basically but you can actually see it working and it's clicking things.
2:46:30
Um and so if they're like there's also the value of like the training data.
2:46:35
If you're getting people using all these websites, you have all this training data of like, okay, they clicked on the blue button, they clicked on the green button, they saw this, they they they entered, this is how they dealt with this form, this is how they dealt with that form.
2:46:46
And so that feels like very very valuable data if you can get it.
2:46:50
So it's probably worth duking it out even if it doesn't uh even if even if it takes a long time. Um, >> for sure.
2:46:56
>> I do wonder where where else they will um where they will plug in.
2:47:00
Like clearly operates at like a higher level of abstraction with like the screen scraping and I wonder if we'll hear rumbles about either perplexity or open AI thinking about like moving up the stack to that level. I'm not exactly sure. [Music]