0:00
Welcome to the new semi-analysis uniform.
Search this video
0:03
This is what we wear every day in the New York City office.
0:06
I come in, I clock in, I put on the Jensen rap shirt.
0:12
They They're just apparently they're sold out or sold out everywhere in Taiwan, but it's just like, come on.
0:14
You got them in this ghost town. >> Amazing. Shout out Timothy. Thank you, buddy.
0:28
Hello everyone, welcome back to SemiAnalysis Weekly.
0:30
I'm joined this week by Doug O'Loughlin, who for some reason hasn't been on the podcast for a little while.
0:35
Doug, what's going on, man?
0:38
>> I thought I wasn't allowed to come on in here, but apparently that was that was never the case. I don't know.
0:42
I've just been I've been doing a lot of things here at SemiAnalysis.
0:46
Um and it's been you know, it's been a crazy season. And yeah, so.
0:51
>> Doug has been busy, but he's back.
0:51
Yeah, today we're going to talk a little bit about the drawdown that's going on in stocks right now, memory optics, lots of AI names are down.
0:59
We're going to um either fan the flames or provide people with some uh >> soothing positivity. >> positivity. TBD on that one.
1:12
And we're going to basically juxtapose that uh drawdown with some of what's going on in the labs, like Sam Altman warning staff that their new pre-trained Doug is going to be RSI in 6 months.
1:23
Um and I'm just continuing to build a ton of models.
1:25
I mean, Anthropic as well.
1:28
Rumors of uh new versions of their models coming out.
1:32
We saw Opus 5 get released, GPT-3, get Doug's reaction to that.
1:34
And some moves around the industry with people leaving.
1:38
So, okay, let's uh start with the lay of the land, man.
1:42
Um stocks drawdown, like what's going on?
1:47
>> So, let's just talk about So, I mean, okay.
1:50
I've been doing this I've been doing this spiel just a wee bit.
1:51
Um so, pretty much up until very recently, like let's say by June 30th, uh at the end of Q2, uh it was pretty much the best best performance in semiconductor history.
2:04
And we've had a little bit of an unwind since then.
2:05
And what you can say is a lot of it you can argue is quote unquote technical, meaning there's like reasons cuz of factors and people were maybe over-levered and stuff like that.
2:15
But the reality is it's like what goes up really really fast often has a little bit of gravity and we're just kind of paying for the massive almost crazy momentum rally we've had.
2:23
But it's gone a little sloppy and it's been very interesting the tape recently.
2:27
Um South Korea is pretty [ __ ] crazy.
2:30
Every single day the stock markets essentially hit limit like they literally limit.
2:36
There's a tweet right now on um where it's like how do I even do my job?
2:38
The the head of HR lost everything and everyone's super depressed and like cuz their stocks are all down.
2:44
Um and it's it's it's kind of just kind of crazy.
2:46
Uh if you look back in the history of Asian financial markets this ironically seems to happen a lot more often than you think. Um like here.
2:54
So like probably one of my favorite books of all time is the great Taiwanese stock bubble.
2:59
I think Taiwan on a per capita basis had like a 100x bubble.
3:01
Like just just incredible numbers.
3:03
Like just everything went to like a thousand times people like, you know, you're talking about like a bank trading 500 times price earnings.
3:09
Like it's just crazy stuff everywhere. >> When was this?
3:13
>> Uh it was in the it was in the late '80s. >> Okay.
3:16
So are you comparing the current situation in Korea to the late '80s in Taiwan or fives are similar? >> Fives are similar.
3:24
The bubble in Taiwan is is truly historic.
3:27
Like I just don't think anything I don't know if if okay, just put it this way.
3:32
If we were to hit that bubble in Korea today, I think Korea's stock market would be like worth 15 to 20 times more.
3:39
Like it was just it's such a crazy bubble.
3:41
I don't think we should ever be able to possibly recreate that.
3:45
Um like it it just it it just shouldn't happen like if they can go off, I guess.
3:51
But I do think some of the lessons learned and some of the behaviors that happened during that bubble are often in- indicative of this one. >> Yeah.
4:01
>> I don't know what to call a bubble or whatever, but like maybe this crazy run-off. What happens?
4:04
You have these like blow-off tops that get absolutely crushed.
4:08
And at the absolute top, everyone's buying as much as they can. They go super levered. They're buying 2x ETFs.
4:15
People are like literally, you know, getting second mortgages on their house to buy more stocks.
4:18
And this this is actually what happened in Korea.
4:20
And Koreans particularly, for some reason, have this like crazy hit rate of always buying the top of every Like they buy the top of every cycle.
4:28
They were buying banks in 2007. I I'm not joking.
4:29
They're buying They're buying CMBs in 2007.
4:35
They're buying SaaS companies in 2021.
4:37
They have like They have this like 20-year track record of always buying the top.
4:40
And Korea just went ultra mega yolo into itself.
4:44
And I think it's a >> Can you clarify a little bit about what's going on here?
4:47
Like you think the fundamentals are different in this case?
4:52
Like >> So, on the fundamental side, I think um they're good.
4:55
But, I think the problem is it's um you know, things are never as bad as feared and they're never as good as you think they will be.
5:03
And I think what happened SK Hynix missed earnings today, missed consensus.
5:07
Um and the reason why is cuz they shifted more to LTAs.
5:08
Uh it's ironic because they were in the United States and when they were doing the ADR, they were kind of dumping on Micron for doing LTAs and effectively getting a lower price.
5:18
But, one of the things that is happening in this memory cycle is like, you know, memory prices, say DRAM and even NAND, right?
5:24
Has probably 3x'd over last year.
5:26
And so, next year, they're not going to 3x again.
5:31
They're going to go up 30 to 50% or something like that.
5:32
But, um finance brains are just absolutely broken.
5:34
It's all about rate of change.
5:37
And historically in the memory cycle, when you have that rate of change go When the second derivative goes down, it's usually the end because it doesn't just go up to a 30% price increase.
5:45
so you should go to -50% price increases.
5:47
So, what happens is they invest all this capital, all the bits come online, and then once upon a time they're like, "Dude, these factories, we can't even run them hot enough because we're running out of chips. There's so much demand."
5:58
And then they build another factory and they're like, "Dude, why why was there so much demand then?
6:01
There's so little demand now."
6:02
They were doubly or they were people were double, triple ordering in order to get their orders in.
6:08
And so, this is like a typical tail whip in a bull whip in in in a cyclical thing, but I do we do think that this cycle is bigger and longer and stronger than past.
6:17
So, I think what's happening is people are looking out and the price increases that the memory companies are showing in in terms of like future price increases because of the LTAs are lower and more conservative than what the market had hoped for and they were like euphoric.
6:32
They were, you know, way over their skis at the top. They're levered up.
6:35
And so, pretty much it's not as good as your dream in your mind could be and you're levered up up 3x and the stocks are against you down now they're down two or three times, right?
6:43
If you're if you're two times levered and your stocks go down 50%, you don't you don't have any more stocks, right?
6:50
And so, many people were at that level though.
6:51
And yeah, and and like so, the entire Cospi right now is down 40%.
6:56
So, if you're over 2x levered, you're completely wiped you're completely wiped out.
6:59
And so, we're seeing just crazy margin calls and then that kind of like that kind of becomes a self-fulfilling thing where everyone who has a lot of money on the table looks at it and says, "Damn, this pile of money is shrinking.
7:11
Maybe I should just sell."
7:11
And then then that like exacerbates it.
7:13
And part of this too is like, "Dude, we have like the the run-up to this was like truly historic."
7:19
And I don't know, being cautious in any way during that period of time, everyone like pretty much laughed at you.
7:24
And now, um I don't know.
7:26
Now, people are going to probably freak out and they'll overshoot on the downside.
7:29
This is this is what markets are all about.
7:32
>> Yeah, do you think there's some smart money on the sidelines right now that is still looking at this and going, "Demand is strong.
7:37
These LTAs are going to continue.
7:39
These are really healthy businesses that are completely sold out for the next number of years and it's going to take a little while to bring more production online anyway.
7:48
>> I think so, but I think the reality is now there's going to be this overhang of okay, well price increases are not going to go up as much as they did and then that multiple that they capitalize it is going to be lower.
7:58
So I'm not saying it's over.
8:00
I don't think it's over, but I do think when you have that big boom bust that usually takes a little bit of time to retrace out.
8:06
>> That or it's going straight back.
8:07
>> But like like okay, sorry.
8:07
Actually let's bring this all the way back to the Taiwanese stock bubble.
8:11
Um I was up late last night um and I was literally looking at stock charts of the great Taiwanese stock stock bubble and during the great Taiwanese stock bubble you had 20 you Yeah, as as you do as a normal person, you had two 40% retracements.
8:24
So it's not unheard of having this giant retracement and then it go continues to go higher, but um it's I mean it's it's pretty dramatic and there's a lot of pain right now.
8:36
So >> Well okay, let's talk about the the people who are um uh calling top and selling and trying to short right now are they have different approaches to you know calling BS on all of the demand.
8:50
I think we can go through a few of them.
8:53
One that was hot in the news recently was the intro introduction of Chinese memory to the ecosystem CXMT YMTC obviously big IPO um what's your take on Chinese memory coming in to specifically the memory makers um industry right now?
9:15
>> So I think on the Chinese side historically everything they touch gets dumped.
9:19
And that's why people are scared because uh they have an ability to ramp a lot of capacity and even if it's at a worst yield, uh it's not like China's trying to make a 10% gross or it's not like China's trying to make a 50% gross margin.
9:32
They usually historically make 10% gross margins and so they're more than happy to dump to win market share.
9:38
Um, you know, the correct framing for how Chinese companies like compete against each other is really it's provinces competing against each other for GDP output.
9:45
Um, and they're not actually competing against each other for, you know, profit margins or EPS or shareholders, right?
9:51
The shareholder is the government.
9:52
The government incentivizes production and each province kind of competes against each other.
9:58
So CXMT is incentivized to produce.
10:00
Um, but I do think the CXMT LTAs are a little bit like I don't know.
10:06
I think I think they're clearly number four in the market and um it's a shortage, so they're going to make a lot of money in the meantime.
10:11
And Apple, for example, is down to using CXMT memory because Micron, they're like accusing Micron of price gouging, right?
10:19
And so no crying in the casino, Tim Apple.
10:21
Uh, well, I guess he's retired.
10:23
No crying in the casino, Apple.
10:23
The reality is you have to buy it at the market price. Um >> Yeah.
10:28
>> And so I think CXMT might ruin the party, but the reality is there's still just more demand than supply.
10:34
And the real question here is how do we know the demand is going to be strong?
10:39
Because like Okay, now I'm just going to go on a rant, okay?
10:42
So demand and supply, they are always yearning toward each other for the crossover of price.
10:48
But in a new market like this, they're blindly searching.
10:51
We don't know We do know the supply curve.
10:53
The supply curve is relatively easy to understand.
10:57
The demand curve we actually don't know, right?
10:59
We know that coding agents and stuff like that means that there's a lot more demand.
11:02
We know that chatbots and the value of like knowledge work mean means that there's more demand.
11:06
But we don't know if it's 10x more demand, like, you know, maybe DoorDash is like uh it's like 100x more demand, or if it's uh 50% more demand.
11:13
And also supply is getting better at the entire time.
11:15
So supply is just going to ramp.
11:17
It's going to ramp blindly into this curve, no matter what, until one day it meets demand.
11:22
And the question is when you're when you're something super in demand like it is right now, you're sitting there and you're like, "Okay, I want to make a data center today.
11:31
I want a gigawatt cluster.
11:31
If I make a gigawatt cluster, I'm going to land Anthropic tomorrow.
11:36
Um I need my [ __ ] to come on time. What am I going to do?
11:38
I'm going to double order my equipment.
11:40
I'm going to triple order my memory because, hey, if I get memory and it's a little bit late, I could just sell it to someone else.
11:44
It's a shortage everywhere.
11:45
So, during semiconductor cycles on the upside, everyone's double ordering, and so the factory looks at it and is like, "Oh my god, there's so much demand."
11:54
And they almost always are overbuilding for a level demand that doesn't come.
11:57
And then that's And then historically, what actually happens is there's usually some kind of economic wobble, uh like a financial crisis or the Fed freaks out, and then everyone just kind of flips out, and then demand weakens for like a quarter.
12:12
Supply ramps because, you know, you can't stop a factory from coming online. These things take years.
12:15
So, your supply is going to ramp this entire time, and then demand sneezes, and you have a factory that isn't that's like, you know, your factory went from 100% utilization to 50% utilization.
12:27
And the only way you can make money back on your factory is like, you want utilization up, so you just you just cut price.
12:31
And that is essentially the semiconductor market in a in a nutshell.
12:36
And so, the real question that's harder than anything else is, where is demand?
12:40
Cuz, you know, if supply's like, you know, supply can ramp 2x or whatever over 2 years, and we semi-analysis exclusively focuses on that.
12:46
The real question is, where is demand?
12:49
Like, that's the you know, the trillion-dollar question, right?
12:54
Um and we think it's wrong. So.
12:56
>> Yeah, I mean, I have my opinions on that.
12:58
Like, in my view, I think demand is pretty obviously very strong for a very long period of time here.
13:03
I just look at my own internal usage.
13:05
I look at how others are using it.
13:06
And I think to me, if you're calling a future where demand stays flat, does not increase that much, or starts to decrease, you really have to believe that the models are not going to be getting better in the future.
13:20
And I see zero signs of this.
13:23
Like, I only see signs of the opposite.
13:25
And so it's I don't know.
13:28
It seems hard to imagine a future where next year or the year after demand is suddenly caught up to by the supply curve as you're describing.
13:39
Maybe I'm not >> Wait, so I'm going to Can I do Can I do some devil's advocates here because I I do agree with you.
13:42
As a guy who now spends a crazy amount of tokens a week or whatever and I'm like probably hooked on tokens.
13:48
Um the two things that people will say is that the technology gets better at a faster rate that people can use it.
13:54
So for example, let's say um the real killer application for uh for AI happens to be data entry and Kimmy K3 is good enough.
14:03
And effectively we make faster and faster and faster cars and better and better and better products, but the real demand curve that matters gets saturated by a product that we already mastered.
14:12
You can argue this was the internet, right?
14:14
Um the reason why the internet's like, you know, the internet bubble you know, they're like, "Oh my god, well the demand is doubling every 90 days."
14:21
was like one of the commoner frames. It wasn't.
14:22
And then also uh the technology behind making the lanes faster got like literally got 2x or 3x better every single year.
14:29
And so then all of a sudden after everything happened, one strand of fiber became 500,000 times more performant.
14:35
And then they're like, "Wait, I don't think we actually need We can't actually fill this fiber demand."
14:40
Um so that would be like the the pushback, right?
14:42
Uh where it's like, "Okay, the models get, you know, 100 times smarter and, you know, 10 you know, 10 times cheaper and that level of intelligence saturates the big market."
14:53
I think that that's um that's probably the biggest bear case I could think of and I think I'm really wondering if that's probably true for some parts of the economy, right?
15:02
Like we talked about this, right?
15:04
Like you don't really need, you know, you're like, "Center my div."
15:07
You know, like does does does Kimmy need to you know, does does uh Fable need to do that or can can Sonnet do that, right? >> my burrito and stuff.
15:15
No, I think absolutely not, but to me there's still um 100 to 1,000 times more people who are currently not using any of the uh currently good enough models that will learn to use them over time. That's one.
15:30
Number two is I just see a whole variety of other use cases.
15:33
It's like internet was connecting people to the internet, but AI is coding.
15:37
It's also chat and research.
15:40
It's also video generation or image generation or drug discovery or material science.
15:45
I mean, somebody just starts making superconducting elements with AI or doing all sorts of other material science work. >> What's that worth?
15:56
>> Well, I think it's worth a lot of spend on GPUs, for example.
15:58
And I think that in many cases the fact that we don't have 10 startups all working on weather prediction right now for farming or for leisure or for whatever is almost purely because everybody's distracted by coding.
16:14
And meanwhile, coding in and of itself the the claim that uh centering a div is going to run out because everybody's going to have a website for themselves, fair enough.
16:28
But coding also uh represents a whole bunch of other tasks that are so much more uh economically valuable intrinsically than centering a div.
16:36
And you can pursue those in in this I mean, there's rumors of Sam Altman talking about RSI, right? >> Yeah. >> Um, and >> Yeah.
16:49
>> It maybe >> think >> Yeah.
16:51
>> I'm going to push back like cuz and I I don't believe this.
16:53
I I I'm more on your side of the camp. Okay.
16:55
I'm just Okay, so the question is will that still fulfill and satiate and all this like will it push and do all this economic good work and we build all the data centers and plus also like I think the problem is like the chips get faster every year.
17:12
We also get better at using the chips and then the models get faster and the models also get better.
17:16
So you have like four rounds of deflation to just make really good products.
17:20
And so all of a sudden you can look up after 3 years and you're like, "Wow, every person in the world has a, you know, a 8X B200 to do whatever every single token every single day."
17:31
And then we're like, "Ah, we've run >> You walked through those four that you say are deflationary.
17:34
Let me tell you the inflationary ones, right?
17:36
The models get bigger every year.
17:38
The models think more every year.
17:40
More people use the models and more people use the models per day.
17:44
Like like each individual user uses the model.
17:48
Um you can look at this as it's happened, Kimmy just tripled in size and now it can only fit on a B300 or an MI355X single node.
18:00
Um can't run Kimmy K3 on a hopper, right? >> Yeah.
18:05
Yeah, I was going to I was actually going to say that.
18:06
I was actually going to talk about that.
18:07
Yeah, I was actually Do you want to talk about that cuz that's a that's like a spicy take is that like we talked about the Um oh oh sorry, two one last thing because I know we're like whatever we're bantering on the on the demand thing.
18:21
The thing that I think that changed my mind more than anything else truly though was last year in coding agents in Claude 4. 5.
18:26
Because I don't know about you, if there's never been a clearer moment where you hit some level of intelligence on the curve and an entire new market showed up.
18:35
You know what I'm talking about?
18:35
Like I could not do this [ __ ] before and then the day after it came out you could do it and I >> You were the prototype, right?
18:40
Like like there was a time last this time last year when the technical staff at SemiAnalysis call it under 10 people were using coding agents and then sometime in November or December you and Dylan said, "Every single person at the company needs to learn how to use this thing."
18:56
And now we have like 90 users of this.
18:58
So that's 10X right there from 9 to 90.
19:04
>> Yeah, and like I think the thing that's interesting is like some of the work gets better blah blah.
19:06
But anyways, smarter model new capabilities that's the Oh sorry, that was the first part.
19:11
The second part, do you use more tokens?
19:13
>> Sorry, from the time when you started to use it in November versus the time today, do you use more tokens than you used then? Per user. >> On average, yes. On average, yes.
19:24
>> I think like 10 times more tokens than you were initially, right?
19:27
>> I'm unsure because that first week I literally did 14-hour shifts for a little while.
19:31
So, I can't like my my initial coding psychosis was pretty hard.
19:34
Like that was sub-agents.
19:37
Bigger models, maybe do it by spend, not >> Actually, yeah, you know you know you know by spend, but 100%.
19:41
If we do by spend, not even close, right?
19:43
Because sub-agents like okay, would you remember do you remember conversation in Claude code?
19:49
Remember when it first came out, everyone remember that one random ass paper called Gastown, which was like, "Oh, I'm going to create this self-healing whatever agent harness that would do this crap."
19:56
And everyone was like, "This is crack cocaine, but I kind of vibe with it."
19:59
I would argue we were kind of there, but you're like you're seeing sub-agents, tool usage, you know, like agents like you know, agent like you know, multi-agent systems, like the fact that you fan out tools like this, like you're already you're seeing the realized version of this, but I think back then it was impossible to market. >> thinking mode, right?
20:17
They're they're they're doing more.
20:19
So, in our experience, that was 10 times more users almost overnight, and then over a span of like 3-4 months, people probably 10x their individual usage.
20:31
Leading to our company going 100x on AI spend, if not more. >> Yes. Yes.
20:36
>> Um Now, the question is every what >> Every company do this?
20:40
Probably not at our level, but I do think a lot of companies have a lot of work to chop.
20:44
>> Well, your argument is this will happen over time, but it will happen on the cheaper models because they will have budget constraints or something like that. >> Yeah.
20:53
Yeah, something like that.
20:53
And okay, well, actually so, sorry. Okay, two things.
20:58
Let's talk about the H100 thing because I think that's really interesting.
21:01
We talk about these new models, it's very clear to me that the old chips will will worthless.
21:06
You know, everyone's like, "Oh, the H100 is an appreciating asset."
21:10
But at some point in time, we're going to be, you know, it's going to take like 100 H100s to to inference one of these models, and you're just like, "Dude, just just let the old girl go.
21:17
Just get a B300 or or, you know, a VR200 instead, right?
21:24
Um that's I'm my speculation, and we'll see.
21:27
We'll obviously be writing in notes about it, but it's like the the true confirmation of this trend is if there is a pricing divergence between B200 and B300.
21:37
>> Okay, I completely disagree with this, but um I'm interested in you flushing it out.
21:41
So, uh the maybe the most fundamental reason I disagree with this is because I don't know anybody who's ripping out H100s to replace them with B300, GB300, or Rubin specifically because the data centers are so completely differently designed.
21:55
So, you would need to justify retiring a Hopper data center because there's no demand for the chips above the price to run them.
22:06
The the the input cost OPEX on energy and people to maintain the data center to justify because you can't mostly for the bulk of the market, you can't replace a Hopper data center in the same footprint with Blackwell or Rubin without like completely ripping things out or just like knocking it down and building new, right?
22:27
>> Yeah, so it's a variable cost.
22:29
>> That is a potential scenario if there's so much demand for the new stuff that people literally just like don't have power permits and and land, and so they just like knock down an old data center >> In a In a frictionless world, that is true, but we live in a world with friction, and the friction is getting worse when it comes to new compute. Is that fair? So, I agree with you.
22:48
>> That's the first thing.
22:48
Now, the the second thing is um the the the Okay.
22:55
So, the camp for GPU prices go down is model progress stalls, roughly, and GPU prices go up if model progress continues, meaning demand continues. That's the rough thing.
23:06
Now, there's one X factor that I don't think is being considered that I want your take on, which is if there is government intervention on the frontier labs.
23:14
To me, if that happens, GPU prices go down, especially the old stuff.
23:23
>> Explain why you think that.
23:25
>> Because they're going to restrict who can have access to the latest and greatest, and that's going to restrict how much demand there can be, and therefore there's going to be more demand for alternative other stuff.
23:36
Um that's going to cause people to use the >> like there's a lot of two what if There's too many what ifs on that one.
23:41
I agree, sort of, but that's like such a top Like okay, so I in my heart of hearts I'm a bottom-up I'm a bottoms-up guy, meaning that the world changes, you know, one mind at a time until it becomes mass consensus, you know, people choosing them at bottoms-up like, you know, product-led growth, right?
24:00
Like this I don't believe in CTOs.
24:02
Like I do believe in CTOs.
24:04
But see >> You don't believe in the great man theory of history, Doug?
24:08
>> We've had this I'm not going to have this conversation. Okay. >> Come on, now.
24:12
>> I've had No, I'm not I've had this one too many with too many what ifs and maybes.
24:15
I'm not going to have this conversation.
24:17
>> in a Jensen Huang shirt, not believing in the great man theory of history.
24:22
>> He He was just a he he's just he's just atoms pushed along by by by the greater Moloch on will.
24:29
Um >> Right place, right right place, right time, eh? >> Yeah, a little bit.
24:32
I mean, I you know, I'm going to be pragmatic.
24:35
I feel like there's many variables, you know?
24:36
It's hard to say that the only variable was him. Right?
24:41
Like I'm sure there's some great guys who who are like, you know, I just think it's um In in my experience In my experience, uh really great horses make the rider look better.
24:51
Now, I will admit, Jensen's a hell of a rider, and he had a hell of a horse.
24:54
So, like, you know, it's very hard Anyways, >> This also doesn't make sense to me because you're you're implying that government intervention is a single person as opposed to a groundswell of minds, whereas I pretty fundamentally believe right now that a lot of people in the US really don't like AI.
25:15
And I think that is not being priced into the >> not priced in. I agree.
25:18
We uh and and how how this will be priced in, I think unfortunately is the midterms.
25:25
Um I'm I'm you know, so this is the the vibe so we we get we get a little vibe uh like we can do a little bit of wind testing in about 2 months from now, and I think um so my guess is it is not a top five priority.
25:36
Oh, no, no, it's not a top three priority, but it is a top five. Is that fair?
25:42
Healthcare >> For for most of the candidates?
25:45
>> For most of the candidates, yeah. >> Yeah.
25:46
>> Everyone's going to talk about it, but no one is going to be like no one's going to platform on it. Does that make sense?
25:53
And I think that that's where Okay, so usually what happens is if if someone if no one platforms on it, that's where corporate interest went.
26:02
>> Uh yeah, and corporate interests are currently stopping, for example, the ROSA bill in the Senate, which uh you know, passed the house like 300 to 20, and yet is stuck in the Senate while people lobby against it. So, >> Yeah.
26:20
>> for those who don't know ROSA, the Remote Access Security Act or something like that, that would restrict China from remotely accessing the Frontier GPUs, um like an Nvidia GB300.
26:29
Uh they passed that bill in the house, it's stuck in the Senate right now.
26:33
So, like if if you're saying if it's not a top three priority, then the lobbyists win over the will of the people {slash} the interests of the candidates.
26:46
>> Yeah, pretty much because dollars >> fourth or fifth priority to number three, where it gets some attention?
26:52
>> How the [ __ ] know, man.
26:52
I mean, you tell me, man.
26:54
I'm pretty interested because speaking of which, I think we literally just started this commissioning today.
26:59
We're going to do survey work specifically focus on this problem because I think it's interesting.
27:03
Like I think personally, I hope you I hope you get a nice little survey in the mail and you you fill it out.
27:09
>> a lot of I think So, here's my view.
27:09
I I don't think that that the >> Yeah, I mean, I haven't really thought this through, but um to me the let's say that the um concept of AI is not forefront in people's mind, but it's used as a scapegoat to excuse other things that people care about like you mentioned healthcare, like you mentioned people's financing or housing.
27:33
A lot of people who have >> climate change >> literally all of the stuff that people actually do care about.
27:40
I think there's a big chance that AI {slash} tech in general or tech people who a lot of people have watched back Trump financially will be scapegoated for other issues that people care about like inflation and the economy or like um yeah.
28:04
>> 100% 100% It is a It is a It is a whipping boy for what you actually care about.
28:09
You know it's important so you have to attach it.
28:12
>> Yeah, I actually I think I disagree with you that I think probably cost of living is the number one issue in a lot of people's mind and that's what they're going to vote for. >> Yeah, number one.
28:19
>> And that AI will be like a sub sub agent >> component of sub agent of the cost of living debate. >> Yeah. >> Yeah.
28:27
>> It's not going to be like we should decide whether we sponsor AI and and push it forward.
28:31
It's going to be like I care about the economy and you should blame tech bros and AI.
28:34
It's probably going to be a big part of the midterms, I think.
28:40
Like there I think the >> We'll see.
28:41
I mean, I'm I think >> tech bros and AI.
28:46
>> Okay, I agree, but I guess I'm just curious to see what is going to really resonate and stick.
28:52
Because there's also like, you know, the horseshoe theory.
28:53
You go super far on the right, you come to conservation, right?
28:57
Like it's like, "Oh, my my ranch is being, you know, the EMF, you know, and the noise from the data center is killing my calves in the farm next door."
29:05
So, I'm pretty curious on this one.
29:07
I'm going to be honest with you, I don't have a strong enough view to be like, "Yeah, this is definitely what will happen."
29:11
Um usually I'm like, "I'll wait and see" kind of guy, but we're doing a lot more work in this.
29:16
Because Okay, so so now I'm going to bring up the last part.
29:20
The last part of what I what would actually kill it is a slowdown, midterms, whatever, stuff becomes illegal, we regress a little bit.
29:25
But I'll also the thing I keep thinking about here is the problem is the future for all these technology booms always comes true, but it's the timing of cash flows that is the issue, right?
29:37
You spend a trillion dollars to get 10 a hundred billion dollars, and it actually does become a trillion dollars one day.
29:43
But it comes five five years later, and at that point in time you're like, "Dude, I don't have enough money to keep this thing going." Does it make sense?
29:50
Like I think that that's going to be the real the real issue is that like, "Okay, you spent two trillion dollars, and all of a sudden, you know, AI rips. AI's at 300, 400."
30:02
We'll say we We'll say we're sending it five trillion, and AI's 500 billion dollars revenue.
30:08
And you'll be like >> OpenAI and Anthropic believes in the final pre-train coming soon because they're going to IPO and have the final funding round. >> Yes.
30:20
And then, when that happens, the final pre-train, effectively, whatever, the final final pre-train happens, it is really good.
30:27
It's the best it's ever been.
30:28
It makes a ton of revenue.
30:28
It grows very quickly, but it doesn't grow at a rate that's enough to pay the bills.
30:34
So, you built a house that you cannot pay for.
30:38
And so, you're paying for five trillion dollars of investment on a 500 billion dollar thing, and you're like, wait, wait, that's 10 years of spending.
30:46
>> You're saying this and that's revenue, not profit.
30:47
You're saying this in spite of your, I think, pretty deep understanding of these companies' financials and how profitable serving the existing models is right now.
30:56
>> Yes, but this only comes So, I don't think we're there yet.
30:59
Like this is like a chicken thing, right?
31:00
Like where it's like, okay, so the thing that matters is like how much. We don't have enough. It isn't enough.
31:07
Like we're still on the narrow path, I think.
31:08
Like where you can kind of blink and see where the revenue comes and it probably is good enough to fit it and you also have these hyper scalers who are super good for the money because they have these other businesses that gush cash and are like the most his like the most profitable in the history of time, whatever.
31:25
Um, but it just like kind of keeps hitting you know, we talk about like blindly scaling the the supply-demand wall, right?
31:32
Um, you know, like I think like, okay, legitimately we think it's like what? We'll say 150.
31:37
I'm going to do round number.
31:38
150 ARR for the entire ecosystem, right?
31:41
Um, and we'll say 1 trillion of 1 trillion of CAPEX, we'll say so far.
31:48
Maybe it's not all in the ground, but 1 trillion of dollars have left the door and there's like a lot of it working.
31:53
So, that's a 15% return on rev, but not on profit.
31:56
So, if we say it's 50% profit, that's an you know, 7 and a half.
32:00
That's not That's not the end of the time.
32:02
That's better than the rate of Maybe that's better than it cost for people to do, but that's not super profitable.
32:06
So, you have to believe that the 150 becomes 500, which is doable.
32:10
And so, that can probably get you to the 500 can get you to like whatever, two, three trillion out the door.
32:17
But then what happens is there's going to be a doubling where just it's really hard for it to double that quickly.
32:22
And that's where I think the gap happens where you're like, okay, we just spent 5 trillion. Yeah.
32:28
>> Yeah, what do you think stops it from doubling?
32:29
I think I've made the case that I I think there would be the amount of demand from all the users and all of the different model types and whatever around the world, but there's also like >> You have to get your grandma to be vibe coding. >> I'll give it a go. I think she's up for it.
32:48
>> It's like grandma, I need you to make 12 agents.
32:51
And so like so so I guess >> I got a really incredible newsletter every month, man.
32:55
She writes it all of her, you know, yeah, I can totally imagine the research being some AI stuff in there. Help with the design.
33:05
>> So so so I think that that's probably the biggest mismatch between and like I don't have a strong view right now, I think and I think we're nowhere near on the path where it's actually like okay, because the thing is the path narrows as you have more revenue and more capex.
33:21
It's it becomes a tighter path to walk. It is not a narrow path.
33:24
There's a lot of cushion, I think right now.
33:26
There's a lot of companies make a [ __ ] ton of money like meta makes a crap ton of money.
33:30
Yeah, their free cash flow goes negative, but if they they wanted to they could stop capex tomorrow and then the profit would print, right?
33:34
That's that's okay, right?
33:36
And so that path is not super narrow, but as you do more and more and more you commit more and more and more the stakes become higher and higher and then the path becomes more narrow and that's what I think would like and then in that narrow path you have to essentially demand people to use it.
33:51
Like people have to be using it like yesterday, right?
33:52
And I think the problem is the decision makers and the people who actually adopt it are two very different worlds, right?
33:58
And that might you know, Zuck's sitting here is like of course you're going to have, you know, your meta sunglasses and you're have your like your your meta wearables and you're going to be in the in the whatchamacallit universe in the metaverse.
34:11
You're going to be using quadrillion tokens every single day and then there's like a grandma in Nebraska who's like Honey, I don't really know how to get my new iPhone to work, you know, and it's like you have to win every single consumer tomorrow.
34:22
And I think the adoption curve is just it takes time and so I think the benefits though of the adoption curve so far is the internet is a really scalable distribution platform that really fits AI super well.
34:34
And so most young people are pretty big adopters.
34:38
Most You know, most working age people are big adopters.
34:41
Like I'm talking to people who are at funds who are in their 40s and 50s who use it every single day because they have to.
34:47
It's a really good technology. >> Yeah.
34:51
>> But the question is will every single person use it?
34:53
And are they going to all be token maxing?
34:56
Because I think you have to get you have to believe that.
34:59
>> Yeah, and I >> I do and I think we both made the case on opposing sides here for a little bit.
35:04
Maybe the more interesting question is on the supply side.
35:06
Like the concept of being able to bring on enough GPUs or hire enough people in order to go and sell cuz like to me revenue probably doesn't necessarily I mean, consumer is such a small part of the revenue of these coding agent companies in terms of AR right now that I actually don't think it depends on the grandmas.
35:26
I think it depends a lot on enterprise businesses and like defense agencies and intelligence agencies and like all sorts of other federal government agencies around the world actually adopting this stuff at scale.
35:38
And uh you know, I I see a pathway to getting everybody and every bank and every Telco and every retail company actually using this in their day job at work um as opposed to like having every consumer have a subscription to this sort of stuff.
35:53
But maybe the more interesting thing is like you're going to run out of gas on the ability to bring GPUs online or to hire enough people in your like B2C go-to-market sales motion or have them install enough GPUs in their private data centers actually run this stuff so that you can get to the point where you can actually grow revenue to that point where you're saying what?
36:13
100 billion AR goes to 5 billion 500 billion AR >> Yeah.
36:18
>> in a year or two years?
36:19
Like >> I mean, we're running we're Yeah, we're hitting we're hitting like some physical like so some of the constraints that are not just like, you know, making the chips out of the factory.
36:27
They're like relatively creative.
36:28
We're running out of electricians in the United States.
36:30
We have like a 100,000 person gap.
36:32
Every electrician who is a mid-journey like let's say a I forgot the name of it.
36:38
It's like essentially like a like a work like a works man or whatever like essentially a mid-level electrician is able to make $250,000 a year pretty easily and if they really wanted to and they worked 18-hour days, I'm sure they can make $400, $500,000 a year and these are people, you know, it's a trade.
36:53
Um these guys are in demand, dog.
36:55
Like they're so There's actually a really weird website that I go to.
36:58
I'm not going to leak all of our alpha on this podcast, but there's a really good website that shows where it's essentially it shows open jobs for electricians and it's crazy because you can do a Wayback Machine and you can see the hourly rate go from like $15, $20 an hour to $50, $100, $200 an hour.
37:17
So these guys are, you know, Yeah, and and the reality is like it takes time.
37:21
You do You don't just like wake up one day and you're an electrician.
37:23
It's like a year It's like let's say 18 months of training and, you know, another doubling requires an entire like, you know, We've never trained that many electricians, right?
37:31
And so another example is capital, which I think is probably the most top-of-mind one for the finance community is like debt, right?
37:39
Let's say so far year-to-date they've the hyperscalers have raised $450 billion ish of debt.
37:46
That's like the biggest ever effectively.
37:47
It's like the third after the United States government, China, and its hyperscalers in aggregate raising capital.
37:52
And the problem is it's like there is a limited supply of money.
37:54
Like someone is buying the debt, right?
37:56
When you issue a bond, someone buys it on the other side.
38:00
And so in order for people to buy more bonds, they have to give them a higher rate.
38:03
That's so you're hitting a supply demand curve there.
38:05
And the demand side, I really don't know what drives it.
38:09
Well, I have a I have a vibe as to what drives it.
38:12
It's like life insurance, okay? Like retiree assets.
38:15
You have to literally And And there's like a weird poetic justice that like the peak buying of annuities is right before retirement.
38:21
All the boomers are in retirement, so the actual the asset classes like larger. But can it double? Can it triple? I don't think so.
38:29
And so that's going to be kind of one of the problems that we start to see is that in order for for um the hyperscalers to raise more debt, they have to do more interest rates.
38:36
And that's going to hurt um you know, when you're a lender and you're saying, "Hey, I can give someone money.
38:41
I can either buy these mortgage-backed back securities for an American, you know, backed by the United States government at a you know, let's say 5% rate for your mortgage."
38:50
I I think that's actually too low. Whatever. For your mortgage.
38:52
Or I could lend to to this hyperscaler who's a better lender than the United States government cuz they make a ton of money and they pay me 7 or 8%.
39:00
So if they keep doing that and all the money goes this way, essentially mortgage prices will start to increase.
39:05
And so you have all these ways that like the system just can't handle another doubling.
39:09
Like the scaling laws are scaling and they're like, "Great, let's make two times bigger model."
39:12
But like I don't think everything can scale at a two or three times bigger rate.
39:16
So that's um electricians and and capital are two two of the ones that I think are creative and weird.
39:22
Um But but I think I think if you give it time, it will though.
39:26
Like they're going to issue a ton of money.
39:27
I'm sure they can issue a trillion dollars next year. Like truly.
39:30
>> I'm I'm I'm actually fascinated by the I I really had not thought that through, but it totally is like pension plans that are funding this buildout right now.
39:41
>> Yeah, but pensions dude, pensions are a secular decliner.
39:42
Pensions as a concept have actually gotten down over time.
39:46
It's not like people are working, you know, people are working more in local governments than they used to be.
39:50
Pensions are And also most pensions are underwater.
39:52
Um so historically the concept of a pension has shifted over to like an equity participation plan, like a 401k, right?
39:59
Pensions have lost share at the expense of 401ks.
40:02
And 401ks aren't really buying They are buying stocks in this stuff, but like you know, they're not really issuing equity.
40:07
So like the really big dollars that get issued every single year, it's through debt.
40:12
And so um yeah, like life insurance is a great great source and the life insurance is from annuities and annuities get purchased when uh uh someone is about to go into retirement. So that's a a source.
40:23
You know, just general insurance actually, you know, health insurance, stuff like that.
40:28
You know, essentially you get all these premiums every year, you reinvest it in capital and that's how you get So it's but like you have to literally believe everyone just needs twice as much insurance.
40:35
You're like, "Dude, no one's going to be You know, you know what I mean?
40:38
Like it's just it that just that just doesn't make sense to me.
40:41
>> Well, I mean it's an interesting balance because there's a lot There is a lot of conversation in you know, advanced economies that there's an inversion in the um uh age pyramid as many people live longer and retire and are building up this wealth for retirement.
40:56
Um but there's a lot of money there.
41:01
>> Less people >> spend it on data centers.
41:02
They can spend it on data centers, too.
41:06
And then and then importantly, the the boomers who, you know, left such a bad legacy >> okay, boomers, right? You can downsize. Sell the house, right? Take that cash.
41:17
You want to buy some life insurance policies and those are going to fund the data center build-out.
41:21
So we're trading houses for data centers, right?
41:25
>> And even better, it takes away your job at the end, too.
41:28
When it So yeah, you can't even own a house.
41:30
You now poor you're poor and penniless and you're in the permanent underclass because your boomer parents invested in life insurance and took away >> But sorry, no, but the life insurance is also funding the data centers which are also building the AIs and the robots that are making sure you don't have a job, either.
41:47
So now you don't have a house, you don't have a job.
41:50
Is it going to come for like retail food service soon as well? Like now you can't eat.
41:54
>> Yeah, dude, that's when the robots come, bro.
41:55
No, I mean that's that's being a little trite, but but yeah, I don't know. It's pretty interesting.
42:00
You just can't say like Um yeah, okay, a a good example.
42:03
Hey, another good example of this actually, my favorite one that kind of makes a lot of sense is Taiwan.
42:10
Like you know, we will just like you know, we can double like we we've been able to just double the output of Taiwan over and over and over.
42:16
And there are economies of scale, but like, you know, I think like 20 I mean, you Taiwan GDP is up like 25% this year just because like TSMC's just like cooking chips, right?
42:25
But, if TSMC was to double again, and let's say there's some ratio of workers that needs to happen, like you're going to run out of people in Taiwan to make the the the freaking chips. Like, I'm not joking.
42:37
It employs almost directly and indirectly, I think it's 20% of the economy.
42:42
So, like the rest of it is like healthcare, retail, and the government.
42:47
There's really only one game in town.
42:49
And so, it's like, okay, what happens if let's just say let's say that they double >> supporting TSMC though, indirectly, right?
42:54
>> Yeah, it's all it yeah, it's all for supporting TSMC indirectly, yeah.
42:55
So, it's like, let's just say TSMC double triples triples, okay?
42:58
Triples the needs of workers.
43:00
It's like, what are we going to like Taiwan needs to literally have more babies to be able to like pay for that future, you know?
43:07
Um So, yeah, just it's kind of crazy if you think about it.
43:11
>> Yeah, I I did just look it up.
43:11
To be clear though, I mean you're saying indirectly, I think that's doing a lot of work.
43:16
TSMC has less than 100,000 employees, and Taiwan has over 20 million people.
43:20
So, it they can they can double it one more time and find some people to employ directly.
43:25
>> They can they can do I'm I'm I'm I'm also knocking on the like the factories that support the factories support the factories.
43:32
Like the material services, all the stuff like that.
43:37
>> Yeah, I mean >> at the percentage of GDP.
43:39
>> They can build a new one in Arizona, right?
43:40
And expand other things globally.
43:42
Like they're they're >> no electricians in Arizona, bro.
43:46
>> TSMC depends on the >> Running out of labor.
43:48
>> I think it's the people that operate the machines. Litho machines, anyway.
43:52
>> There's there's a there's there's actually a crazy thing.
43:53
Well, to make a new fab, you you need an electrician.
43:55
Um the crazy thing there's like um people are flying electricians via Cessna to like like, you know, to like little backwater places in order to like make it to the job sites and stuff.
44:08
There's a private There's private flights just filled with like 16 dudes of electricians like, "Ah, you're you're here for your second shift." >> So, okay.
44:15
So, this this really reminds me of the oil fields in Canada.
44:17
I have some friends who have worked on uh different oil projects Fort Mac and stuff like that. Um >> Yeah.
44:26
>> the the boom bust cycle of oil and like Calgary and northern Alberta and stuff is really fascinating as some of these communities kind of get destroyed as like you know, uh oil mining whatever else is uh like, you know, kind of shut down.
44:40
Um it's going to be fascinating to see how that happens where it's not even a natural resource that is drawing people to these areas.
44:47
It's literally just like a permitting >> An information An information factory. >> Yeah.
44:55
No, but the the reason why they have these data centers here is because of the ability to produce energy which like obviously being close to the energy sources is good, but generally it's just like they put up data centers where they can get a permit most easily.
45:06
And uh that's that's where they put the >> West Texas. >> Yeah, exactly. Um >> So.
45:12
>> So, it it but it's it's not even like there's something fundamental where um you can just run out of lithium to mine and then or gold and then all of a sudden everybody has to turn around and go home.
45:23
Uh it's literally just like if a permitting regime changes, like if the political will of the people changes a little bit, then all of a sudden everybody in there is out of a job and things change a lot.
45:35
I mean, I think we saw exactly this when we were putting out stuff on New Mexico.
45:38
We had Ali on a couple weeks ago to talk about that with Jeremy. >> Yeah.
45:43
>> Um and the whole >> But the wait, the thing is okay.
45:44
So, there is there, you know, like the the invisible hand of capitalism does work a little bit.
45:49
There's just also this interesting aspect where you think about it is if they un- So, if they do this over and over and over, the logical conclusion is all these new jobs just get disappeared and then some guy is going to be like, "Whoa, they took our data centers. They took our jobs."
46:02
And so like there is a self-correcting force in this as well.
46:06
And cuz like And this is an interesting poll that I saw the other day is like people who hate data centers often don't live near one.
46:13
While people after having a data center in your community, especially for young people, it's a net favorable thing because jobs.
46:22
>> Yeah, I mean I went to visit one in the Buffalo area and just like everybody we met I mean of course they they like having a job but everybody was just super pro the whole site.
46:31
How much it's employed >> Cuz they're just like cuz they're like, "F yeah, dude. I got a job, dog.
46:37
A construction job is good."
46:39
And like I you know, it doesn't take like super high-skilled labor.
46:41
You don't have to have a PhD. You know what I mean?
46:44
Like it's just like, "No, dude.
46:45
It's like it's an honest living, okay?"
46:47
And you're making I mean you're people are making more money than they have been because it's clearly in demand.
46:50
I mean it's just you say supply and demand.
46:54
When there's a lot of demand, supply's got to reach it.
46:56
And so it's like I think it's a net good thing if it's broad.
47:00
And I think ironically because these data centers are being built in the middle of nowhere, it's actually very good for broad participation, right?
47:06
Like you can argue the entire finance industrial complex and the technology industrial complex is extremely narrow. It's like, "Yeah, dude.
47:13
San Francisco took, you know, is is is making me making an app so I can deliver something in a rural in like a rural state."
47:20
That that is not broad participation in the economy.
47:22
But building a building that is you know, a $10 billion in the middle of nowhere that has extremely high needs of electricity, high levels of service, extremely important just like you know, you know, people are willing to pay for overages because it's extremely you know, important for the strategy of these companies.
47:40
That's broad participation cuz you you know, a thousand people I I don't know what the math is.
47:45
I I feel like Jeremy knows.
47:46
I I want to say it's like 10,000 people for gigawatt or something like that.
47:49
And so it's like, "Hey, if you bring 70 gigawatts, it's like you know, it's it's a lot of jobs. It's 700,000 jobs."
47:54
And that that starts to move the needle, dude.
47:55
If you do a you know if you do a >> In terms of revenue, right?
47:59
There's many other jobs programs where you can spend less money and create a lot more jobs, but a lot of them are like not sustainable in the same way where this is obviously going to run for a very long time.
48:10
People really like these data centers.
48:12
And I think that um generally speaking a lot of people have problems with jobs programs when compared to investment from companies that happens to be directed to these communities that happens to result in jobs. Um in investment.
48:26
>> It's more sustainable.
48:26
It's more sustainable, so.
48:28
>> Yeah, it's good stuff.
48:28
Okay, we got to move to a wrap here. Yeah. >> Okay, yeah.
48:32
I was going to say I have a I have a call I'm late for, so.
48:33
It's the reason why I'm not on this on this podcast, bro. >> Good to have Doug on.
48:37
Good to debate a little bit. Yeah.
48:40
Good to uh get some hot takes.
48:43
>> Well, some some modest takes, I think.
48:45
Oh, we should get Joey on.
48:45
I really want to do a finance week earnings recap. >> Okay.
48:49
We'll have to get Joey on. >> can shoot the breeze.
48:52
>> Or we'll have to get Joey a a competitively dynamic shirt as your Jensen Huang lightning leather jacket shirt.
49:00
But uh >> We have a Morris Chang.
49:02
We have a Morris Chang shirt, so.
49:06
>> Okay, Joey can get a Morris Chang for that.
49:08
>> We can get a Lisa Su in here, too. Be sick.
49:10
Everyone This is We actually have a We have a we have actually new company uniforms, so. You're looking at them. >> Okay, sweet.
49:19
Thanks for Thanks for coming on, man. Good job. >> Yeah.