0:00
The issue in the physical AI world is diffusion.
Search this video
0:02
If your company could exist two years before you, then that's probably not right.
0:06
There's more diversity among partners within a firm than between the firms. >> Absolutely.
0:11
Do you think we'll see a similar thing that we saw with, for example, search engines back in the day where, you know, nine out of 10 of them just won't exist in five to 10 years?
0:19
>> In the entire history of the company, we've never spent any money we've ever raised. >> What?
0:23
>> What's very important for for for listeners out there to know is you can have much more generalized models that go on to machines.
0:30
>> Carter, thank you for joining us on Giant Ideas. >> Thanks for having me. >> Really appreciate it.
0:33
We've been a giant been investing in physical AI for a while.
0:36
The very simplistically for us, it's like 85% of the global economy is physical.
0:41
It kind of stands to reason that AI can have a much more profound impact and a bigger outcomes when it's applied to the physical world, but it's just way harder.
0:49
We kind of like that cuz it creates barriers to entry.
0:51
It's way harder, in my opinion, to do what you're doing with AI than, you know, building a chatbot or whatever.
0:58
Talk talk to us before we go deep on apply just about why you think physical AI is going to be so big.
1:02
It's probably less well understood generally, but but what's your kind of thesis on physical AI?
1:09
>> Yeah, [snorts] I think what you said is exactly correct.
1:10
You know, I've said before that I think when we look back 25 years from now and we look back at this time like we look back at the beginning of the internet, which is roughly around that time.
1:22
In the beginning of the internet, static websites and the [snorts] companies that are, you know, interesting, you don't even know them.
1:28
You actually don't you know, think about those companies.
1:33
And even the big ones then, the AOLs and the Prodigys, I would say most people in college right now would not be able to they don't even know their names.
1:42
But so who who do we know?
1:42
Well, we know Google, we know Amazon, we know Apple.
1:44
I think when you look back, it's going to be it'll be things that are impacting the average person in their day-to-day existence.
1:54
And the heuristic that I use is, you know, you're sitting at a you know, you're at the airport and you're sitting at the gate and you look around and no matter where you're going, even if it's San Francisco, you can look around and say like, how many of these people, you know, really are like the next, you know, Fable is like on top of their mind.
2:14
Cuz it's in a it feels like when you're in your world of the internet, in your world of of of uh software development AI that everybody knows that Fable is and it's not what you can't use it for these days. And now you can use it.
2:26
It's like people don't even know Claude.
2:28
And I'd see it's and and if that sounds very strange to you as a listener, that just shows you how much of a bubble you're in.
2:36
Like like like if that if that seems really crazy to you, >> Yeah.
2:40
>> actually most people don't know.
2:42
Actually most people don't know these companies.
2:43
They they they they don't they don't care about that SpaceX IPO. >> Right.
2:46
>> Um and so what do those folks do?
2:46
Well, they drive taxi cabs and they work at the local convenience store or they maybe they're a retiree.
2:52
And like how is AI going to impact their It's a very different than kind of this you could say like, you know, uh unembodied AI.
3:03
Like this this digital, you know, this this large language models.
3:06
So I think the embodiment of this of this intelligence, that's where it's going to meet most people.
3:12
>> Do you think that the the big LLM companies, the vibe coding companies, do you do you think we'll see a similar thing that we saw with for example search engines back in the day where, you know, nine out of 10 of them just won't exist in 5 to 10 years? Can you see that?
3:25
Can you see some of these very big LLM companies going to zero?
3:28
>> I don't I don't like making predictions like that because it's it's such it's you know, there is a little bit of fooled by randomness kind of things where if I can say enough things, some of them will be correct.
3:36
But I think but there are patterns.
3:38
There are patterns in markets and um the reality is you do tend to coalesce towards a couple of winners and those folks make so much money that then they get into other things.
3:50
And I thought the question you were going to ask is do you think some of these LLM companies are going to ultimately get into physical AI? And the answer is yes. >> Yeah.
3:57
>> It's it's a natural extension, I think.
3:59
I think that is the next big wave. >> Yeah.
4:01
>> Um and you see kind of the early remnant or the early, let's say the you know the the first photons hitting your your your eyeballs over the horizon. >> Yeah.
4:09
>> It's a open AI in this you know this hardware device that might be coming or not coming.
4:13
And the lawsuit that just got announced you know with Apple suing them.
4:17
So these companies are going to play more and more in the real world.
4:22
I think that'll absolutely happen.
4:23
Anthropic getting into drug design >> Yeah, for example. >> Absolutely.
4:26
So I think I think when you So then you have to kind of put a little like let's say boundary around what is physical AI.
4:33
And they can be many things.
4:36
It can be you know designing nuclear plants in a more efficient way or something like that.
4:40
In our universe, in the Applied Intuition's universe, it's specifically on making machines more intelligent, putting intelligence in the machines that move around us and move move goods.
4:50
The most obvious easiest one is a car. >> Yeah.
4:55
>> Uh but maybe like the non-obvious ones are construction, mining, um drones in defense.
5:00
All these things are physical machines that move.
5:03
And by adding imagine you just take a cup of you know intelligence and you spill them on all those machines.
5:09
Those machines are all going to get way more safe.
5:12
They're going to get more productive.
5:15
Um and they're going to they're going to drive efficiency for for society.
5:16
And I think unlike the AI revolution that's that's that that's creating so much anxiety in for knowledge workers specifically in in things like accounting for example uh or in finance putting intelligence into a combine on a farm is there's no one's resisting that.
5:38
The average farmer in America is 58 years old.
5:43
Their kids are not coming to take over the farm.
5:46
The the so there and the the other kind of crazy status something like less than 10% of farmers are under 35.
5:53
So it's like it's really like there's a there's a there's a big gap opening.
5:57
If you look at Japan, people always talk about the demographic collapse in places like Korea and stuff.
6:03
What they don't really also talk about is the remaining people, what are they working on?
6:05
And in Japan, guess what they're not doing?
6:07
They're not driving trucks.
6:09
Uh they're not they're they're not doing these kind of back-breaking, laborious tasks.
6:13
And so for in the spaces we play in, defense is ex- self-explanatory.
6:20
AI can't get there fast enough. Right?
6:22
So you don't have that resistance and that pushback of of that.
6:26
Um where you see the most of that is is in something like our taxi example.
6:30
Sometimes in trucking, but even in trucking, even in the United States, there's a there's a much larger labor shortage than is anything else. >> Yeah.
6:37
>> Uh but by and large, our domain in physical AI is one where it's like uh you're kind of like a you know, you're cooking a diner and and the people sitting at the table are like, "Is the food coming?"
6:47
because, you know, we're ready we're ready to we're ready to buy.
6:50
>> Well, the the the robotic chef or whatever is one example, but you've got some probably even more life and death scenarios in that, right?
6:55
So if let's if ChatGPT has a lag, if it's a 5 seconds late, it kind of doesn't really matter.
7:01
If it gets something wrong, if it hallucinates, it doesn't really matter.
7:04
But if your AI makes a mistake, it's literally the difference between life and death if it's powering a kind of huge truck that kind of drives off the road or if it's doing a drone that, you know, shoots the wrong thing or whatever, it is life and death.
7:14
So it's way harder the kind of the barrier to entry is way harder.
7:20
Um the stakes are high, basically. >> Yeah.
7:21
>> Tell us a little bit about just uh for non-technical listeners, what is the great challenge of latency, um of accuracy that you've had to get right which has led you to build this huge business where you've got, you know, I guess almost a billion dollars of revenue or whatever, serving I think 90% of the top car makers, serving the US military.
7:39
High-stakes environments.
7:39
What is the technical challenge that you've had to get right?
7:44
>> So, roughly two big areas that we're uh you know, we're building models and have intelligence in.
7:50
Uh one you could simplifying everything.
7:52
Just it's it's tough, but just just do that.
7:54
One is you could say it's off-board.
7:56
Um we generally this is the Sometimes you call this a world model.
8:02
It's this is the environment where you test and deploy this intelligence.
8:04
You figure out okay, uh it's the tools to make the brains which ultimately go on the machine. >> Okay.
8:14
>> So, those are the two that's off the machine like the developer, let's say, you know, IDE. >> Yeah.
8:18
>> And then on on the machine.
8:18
So, off machine that's [snorts] just a that's very much like a like a cursor or a claw, right?
8:26
You're you're you're It's a development environment to make sure that um the models that are going to be get deployed are being effective, efficient, you're testing them. >> Yeah.
8:38
>> Um this is like everything from neural simulation, this is from synthetic data.
8:42
That's that whole class of problems.
8:44
>> So, neural simulation for example, that that is basically not on the actual hardware itself.
8:48
You're This is just you're just simulating things.
8:50
You're simulating thousands, millions of different scenarios to build the intelligence that you can then later put on the machine.
8:58
>> Yeah, we do tens of millions of simulations a week.
8:59
The difference between the first and the second one, which is an imitation-based or like let's say a learned approach, is in the learned approach that environment has to be um dynamic in a way because as the AI moves in the environment, it will bump into agents and and and those agents need to respond to it.
9:16
It's not just testing against a pre-set set of scenarios. That was the old way.
9:23
Million scenarios, can you make them through you know, can you get through them digitally, and then we deploy the model on the vehicle.
9:27
Today, it's like actually let's just let let the model learn how to drive in a virtual environment. Yeah.
9:35
Uh specifically the word you're using was uh, neural simulation.
9:36
That that's a particular sub subcategory of simulation where you can take um, images and essentially recreate a world digitally much faster.
9:46
Historically, the way you'd create a simulated world is like slow and painstaking and very much akin to like how Hollywood does CGI. >> Okay.
9:55
>> You have technical artists that are creating assets and then you're there's material properties and there is a heavy emphasis on a kind of a reality to simulation gap.
10:03
Reinforcement learning and neural sim and that this new world it's it's it's a little different from that. >> Yeah.
10:09
>> You're all in the same broadcast I mean you're zooming out it all is quote unquote simulation or it's all quote unquote world models or whatever, but um, there's a lot of nuance in that.
10:17
That entire category you can just think of it as like the tools you need to get something to be intelligent to make intelligent models.
10:26
So, that's the one big part of the business. >> Yeah.
10:28
>> The other part is what you originally talked about is now you have these models and you're going to put them on these machines.
10:32
Okay, we're talking about lots of different machines here. talking about.
10:34
And when I mean different I mean not only the form factor like a drone versus a tank. >> Yeah.
10:40
>> You're talking about that's on the ground versus in the air.
10:41
You're talking about the compute realities.
10:43
Uh, you're talking about the um, the the the mission.
10:48
The you know, are you going to is this car going to is this truck going to run for 12 hours in long long haul or is it going to move in a factory floor for short amount of time but with lot lots more going on around it.
10:59
So, the models historically or the way that that autonomy was historically built was each of these verticals would have very specific um, you know, models made for it.
11:10
And what's happened in in LLMs is you have this paper this attention paper um, and that create you know, introduces transformers.
11:20
Transformers results in chat GPT and everything like that.
11:24
A very similar thing happened in self-driving.
11:26
Self-driving historically was done differently and post transformers is done differently.
11:29
And um, and today uh, the what's very important for for for listeners out there to know is you can have much more generalized models that go on to machines.
11:42
And so what we're doing at Applied Intuition is that kind of our our you know, in the Peter Thiel way of like what do you get that other people didn't get is that we can actually work across multiple verticals using the same models.
11:55
But you're really doing inference on uh on a chip that's not maybe very beefy because the buyer or the drone or whatever there's constraints there.
12:04
It might be a constraint because the consumer doesn't want to pay a lot for a car.
12:10
So they're not going to pay for a $10,000 $20,000 computer in the back.
12:15
And so suddenly you got to do everything on a $500 chip.
12:17
And that now this this model that works really well with a lot more compute and a lot more sensor data suddenly doesn't work as well. >> Okay.
12:27
>> And so that's kind of that that's our our our other part of the of the world is take mod take a model that works in all these different form factors just as effectively. >> Yeah.
12:36
>> And then you you also said there's so many things we're covering here.
12:38
You know, you also said this thing about chips.
12:40
Like what what's what what is you okay, so you have latency of all those things are kind of comparable regardless cuz these are physical machines and they interact with humans.
12:47
And so you you're always going to have some similarities.
12:51
But on the silicon you're not going to have similarities.
12:53
You're going to have huge huge vastly different realities because historically the intelligence was the humans. >> Yeah.
12:59
>> So a human needs to do different things actually in a dirt mover in a haul in a haul in system than on flying a drone.
13:06
And so therefore the the processing power in all these machines is different.
13:11
I think one of the things that we've done which is particularly special and and does a wee we're hardware agnostic.
13:16
We run on lots of different compute. >> Yeah.
13:20
>> Way easier said than done.
13:20
And if you're a technical in the audience you're there's another 50 questions that have to get asked at that point. >> Yes.
13:28
>> About like how do you abstract hardware away from software in the way that we're talking about and that's another huge part of the business and we won't we won't get into it but but those are the two big areas where you're where we're making the environment and we're making the models that actually run the machines.
13:40
>> Is it a kind of lazy way of describing this as you bring it back to cars, Waymo maybe like a kind of Apple where it's dependent on like their vehicles that they're they're working on and then >> Yeah, the business terms are verticalized. >> They're verticalized.
13:51
>> Yeah, they're doing the chip, they're doing the sensor.
13:52
They're not doing the car but they're augmenting the car so much and and it's it's their consumer brand and it's their app. >> Yes. >> Yes.
13:59
>> Whereas what you guys are doing I think of it a bit like kind of Android in the mobile phone world where it's like it doesn't matter if it's a Samsung phone or Huawei phone or whatever.
14:05
In your case it doesn't matter if it's a, you know, a Fiat car or a drone or a truck made in Japan.
14:11
It's like >> All our customers by the way, yeah. >> Yeah.
14:14
Your intel intelligence has to work on all of that. >> Exactly. >> And okay.
14:18
And um how did you do that?
14:22
>> Yeah, I mean it's uh it's >> [snorts] >> That's the story of the company.
14:24
We started originally in automotive.
14:26
That's the bread and butter that I went to the General Motors Institute undergraduate and I worked as an engineer in the automotive industry and I think that's it has a huge impact influence and impact on on our company.
14:38
Um and fairly early in the company's history like literally in the first year we realized oh actually the problems are the same but it's the self-driving kind of the the the you know the technology didn't exist at the time.
14:49
Um and uh I we entered the self-driving arena when the technology was frankly more ready and I think that was a huge plus for us because again this takeaway for founders is um probably the biggest thing you have to get right is timing.
15:05
I mean the almost everything else maybe your co-founder or or can can you can kind of change and like maybe it's the other thing you can't change your co-founder like what what's co-founder relationships break the company's been put in dire situation.
15:20
Um but outside of those things you can almost change everything.
15:21
You're like, you know, you have some not so good investors, not a big deal.
15:24
You can ignore them more than than others.
15:26
If you need to change the products, you need to change the market, you need to change the company's name, you need to change the All all the all those things are two-way doors.
15:34
You can come in and out of them.
15:36
But but the timing aspect is one of these things that you have to really really focus on and say and heuristics. >> Yeah.
15:45
>> So like if your company could exist 2 years before you >> Mhm.
15:49
>> in the same shape and form then that's probably not right.
15:52
And if it going to exist 2 years later then where that's probably not right.
15:57
So if you it was too early, it's like my the technology doesn't exist to do what we wanted to do.
16:01
If it's too late, there's too many entrants, there's too many players.
16:03
So you have to get in at the right spot where you know, it it's like the technology has just flipped over and you're you're maximizing its use for the thing.
16:11
And so not to be too early, not to be too late. >> Yeah.
16:17
And because [snorts] you as I understand it, you're not actually that big on pivots, are you?
16:20
You you feel Am I right in thinking you think pivots can be quite dangerous once you've raised a bunch of money and hired a bunch of people?
16:26
>> Well, so yeah, a nuance there.
16:27
In the classical YC and I still like pitched the YC book because, you know, that's kind of the university that I went to.
16:34
YC is very pivot friendly.
16:34
It is a belief is you get and and that's because you're looking at the big bang of the startup.
16:39
You're looking at the first milliseconds of the company emerging.
16:42
There's lots of, you know, things happening and I think in that point you should always you should be completely open.
16:50
The moment you raise the a dollar, the moment you hire the first employee, suddenly the universe is now becoming the laws of physics are becoming more defined and those are much more difficult to change.
17:00
And and then you fast forward to you have 50 employees, you have 200 employees.
17:04
There are companies that famously pivoted and done great. Twitter was one of them. But it's more difficult. It's more difficult.
17:13
Just because like you know, you evangelize to employ 34.
17:19
This is a great company and you're still early enough and we're going to do these great things and then, you know, 6 months later they're like, "By the way, that you know, that hill that we're going to take, that's not the hill that we're going to go the other way."
17:29
And uh you know, as a general in a battlefield, you might lose your troops in the process. >> Yeah, yeah.
17:35
>> You, in fact, are backed by three of our most prominent best guests we've had on John and I have is Ray Dalio, Mustafa Suleyman, and Reid Hoffman. >> Yes. Yes.
17:43
And some of that I don't want to also discount my other investors.
17:45
I have to >> No, well, I imagine they're [laughter] a small piece of the pie.
17:48
So, you raised around a billion dollars, is that right? >> Yeah, yeah.
17:51
We have We've We've been around for about 10 years.
17:52
So, you know, and we do kind of regular drumbeats of fundraisers. >> Yes.
17:58
>> But we, you know, you told me that the audience is founders and stuff like that.
18:01
A lot of times founders ask me, "Well, why do you guys raise money?
18:04
You know, we've been famously, you know, we have a business which is like sustainable, which is >> [clears throat] >> another way of saying like it makes more money than it consumes."
18:15
>> [laughter] >> And and that's been the history of the company.
18:18
So, in the entire history of the company, we've never spent any money we've ever raised. >> What?
18:22
>> So, Ray's money and Mustafa's money and whoever, you know, whoever Marc Andreessen from the beginning or or Hemant from General Catalyst, whoever it is, Mamoon from Kleiner, all their money's in some bank account, you know, earning some a lot of interest.
18:36
>> I think Treasury decent decent yield these >> Yeah, actually they You know, once you get to big numbers, that actually also kicks out money.
18:42
But the But the punchline is uh you know, the the uh the the investors actually, whether they're famous like, you know, the folks that you just mentioned or or or or people that are not, I think for founders it's important that you don't you you you're kind of assembling the cap table almost like like if you were building like you're having a dinner and you can invite anybody on the planet.
19:07
You don't want just your friends.
19:07
You also don't want a bunch of people who are just famous and you don't know and are useless.
19:12
It's kind of like a you know a like an intentional way and the theme of our company is always been like intentionality.
19:18
But the more tactical question that founders always ask is well, why do you keep raising if you don't need it?
19:24
And actually it is for their ideas.
19:26
The reason the venture ecosystem exists and the reason that it there isn't venture capital for laundromats is because it scales really effectively.
19:35
So that means the prizes are bigger.
19:38
That means the number one to number two distance is is is is more important.
19:42
Like being number one is really important.
19:45
You can be the fifth best laundromat in town and still be okay.
19:50
You can't be the fifth best uh you know whatever like credit card company.
19:56
You got to be Ramp, you got to be Brex and then it's like dot dot dot.
19:59
There's you know it's it's it you're not going to get the spoils.
20:01
And in order to be number one you're just maybe a couple percentage points in decision-making better.
20:09
Obviously abstractly speaking.
20:09
And then so having the right people around you is important who are financially incentivized to see you succeed.
20:15
We really try to curate the cap tables of the companies that we back at Junction when we lead rounds.
20:20
We really try to think about not just who are the kind of big uh co-invest venture firms, but also who are the angels, how do we surround the people to basically give it the best shot at succeeding.
20:29
And often we found that you know yes, if you can bring in a mega-cap US venture firm, clearly that has a halo effect around the company that helps them attract more capital.
20:40
And I think that's probably even more true today that if you get one of those sort of top five mega-cap venture firms, it's a kind of self-fulfilling prophecy.
20:46
But [snorts] actually often we found it's the angels, it's the former CEOs who can provide the biggest value.
20:50
Like we've got a physical AI company here in the UK called Causaly. >> Yeah.
20:55
>> And any AI coming into that, kind of top US mega-cap firm was was transformational because it really put the halo effect around it.
21:00
But then we also tried to get people like John Brown, the former CEO of BP who was on the Intel board for many years, One of one of our advisory board members go in as an angel and an advisory board member.
21:10
And he's had a massive impact in making intros and helping the CEO overcome some of the corporate LPs, people like Henkel as well, kind of industrial partners who then become customers.
21:20
>> How How do you think >> I have a bunch of thoughts on a couple of those things. >> Tell Tell us.
21:23
Yeah, go >> before you even ask me the question. >> Okay.
21:26
>> [laughter] >> The The The The first thing I would say is, you know, this is my third company and I'm I came out of YC.
21:32
So, I I could pick >> Mark Andreesen to be on the board. >> Right.
21:36
It's It's It's a weird asset class where the, you know, the asset picks the manager.
21:39
Whereas most other asset classes, the manager picks the asset, right? And uh or all, maybe.
21:42
The The But my first companies, you know, we couldn't raise any money if we tried.
21:49
I mean, we tried and definitely we tried.
21:52
Like, we tried for years.
21:52
So, the macro point I would say to founders is, you could probably change some people on our cap table and it doesn't mean that Applied is fundamentally like a bankrupt company then.
22:03
Um But at the same time, uh you know, you hold that into contradiction with what I just said earlier is you want people who are giving you the right ideas cuz that one or two percent or five 10% abstractly speaking kind of going the right direction can [snorts] have an impact in you being the in the pole position versus being number three.
22:19
And that can have a huge impact.
22:22
In terms of um assembling the cap table between operators and and VCs, I think the multi-stage funds are more interchangeable.
22:34
Um if you're not getting the Hemants, you know, the Mamoons, you know, arguably top two or three investors in Silicon Valley, because they've also like there's like a like a language model.
22:46
They've absorbed so They have so much uh experience, seen so many deals in this very obscure edge case world of private venture financing that they can actually incrementally help you.
22:59
And then there's the the old adage of like does a company make money or does money make a company?
23:04
Like if you and and do what you alluded to earlier is today there's more of the money making the company than than maybe five or 10 years ago.
23:13
Then it goes in abs and abs and flows.
23:15
Whereas if there's a consensus bet on a specific company and all the major VCs are in it, they're more likely to win. >> Right.
23:24
>> Like then the the best employees and the best engineers they join the best researchers join and then because there's a bet they make the best product the best product is the best customers the customers then tell the investors and then you suddenly before you know you're like the chosen winner.
23:37
So that's another strategy as a founder have to like kind of keep in mind or try to get after you know that you can just get money.
23:45
>> First is just get any money. Sure.
23:46
>> Then try to get good money.
23:46
Then we talk about corporates.
23:48
This is a very particular wordy.
23:51
My YC experience you know CVCs were like like a profanity. >> Yes.
23:59
>> And uh >> I think that's outdated now, right? >> I don't think so.
24:04
>> You know that's >> That was my my what I was going to push back on.
24:06
I think I think it's kind of like the multi-stage funds actually the more important thing is the person that's that's there.
24:13
In multi-stage fund who the partner is you could have the three founders have the same firm and have wildly different experiences because of the of the partners.
24:21
Cuz the the firms are actually there's more diversity among partners within a firm than between the firms.
24:29
>> So like you you you could you could let's take let's say the top firms you put in Andreessen you put GC you put you know Kleiner.
24:36
Number of those partners could work in other firms and be completely successful and be but you know within Andreessen there's going to be such a huge huge diversity of people.
24:44
Now if you talk about CVCs they're consistent in the sense of the corporate mandates are actually more aligned they tend to be let's invest in something that accelerates the business.
24:56
They tend not to be purely financial.
24:57
They're they're they you know, maybe they're they can be a distribution for for you and that's why they want to get involved.
25:04
The partners themselves are different, but the thing that I see a lot from the YC days and I haven't been proven differently here is the it's one thing is the who's the VC.
25:17
And then there's the corporation.
25:19
Acts very differently from who the CVC is.
25:21
So the person who's in the CVC, they could have the best intentions and they have and everything is great and then the corporate is like decides to get in the product line that you're in or the corporate is running out of money and they need they need you like there's all 50 other things that can happen or the corporate decides that they don't want a CVC anymore.
25:37
And suddenly you're getting a call and they're like, "Hey, our our you know, our fund is shutting down and like we'd like to sell our position."
25:45
And you're like, "Wait, what?
25:45
That doesn't I'm not I'm in a Series A company.
25:48
What what are you talking about?
25:49
You can't you're going to sell the position?"
25:50
And and I've seen at least CVCs have a much higher turnover.
25:56
Like just it's so then you're exposed to the corporation more.
25:57
So I think you have to be of all those categories, operators, the multi-stage funds, the CVC you want to be the most particular on.
26:06
And like you know, I can never be anything but what my and sometimes I get in trouble because I'm just so direct.
26:15
I generally wouldn't take CVC money.
26:17
Yeah, we we have we've taken some.
26:17
We've taken some and I but it's like extra I think it's I mean it's it's a small slice.
26:23
You want to be thoughtful and and we entered those conversations with that with that view.
26:28
I think generally speaking the also for your audience, you know, all of my advice and all of my thinking is in a very particular, let's say, lens.
26:41
It's for like Bay Area deep tech AI companies and I would even say South Bay companies Um, that are venture backed and uh, you know, are are predominately software.
26:54
They're not We're not a hardware company.
26:55
We would dabble in hardware, we're not fundamentally hardware company.
26:58
And so, everything's hyper focused on that. >> Totally get it, yeah.
27:02
>> Yeah, because it's The answer changes very quickly if you're in London. For sure.
27:05
Answer changes very quickly if you're in Tokyo. >> For sure. Yeah.
27:07
No, I I I hear you on the on the corporates.
27:09
I I think definitely we've seen it get better.
27:10
You know, it used to be terrible where they'd have these horrific terms where they had the right of first refusal to buy companies.
27:16
We we've seen it work a bit better recently, but but I think Who Who's been >> But if you don't have if you don't have any choices, I would take all the corporate money on the planet. >> Sure. Sure. Sure. >> You got to survive. >> Yeah, yeah, yeah. You can pick.
27:27
Who Who's [laughter] been Have you found that um, the big venture firms have been more helpful or has it been more some of the the kind of operators that we've talked about?
27:35
Who's been the most helpful to you?
27:37
>> Again, the context is important. I was COO at YC. >> Right.
27:41
>> So, I'm walking out, you know, I'm I'm I'm one of them in that way.
27:43
And then I you know, but before that I'd done multiple companies. I'm an engineer.
27:48
>> So, uh, my kind of ethics is different.
27:50
So, what I think is valuable is different.
27:52
All of them >> Um >> You You look at They're they're all technical.
27:58
Uh, they've all started things themselves.
28:00
Um, these are folks who are You know, they're they're the mix of the operator >> Um >> Uh operator investor.
28:05
So, that you kind of get a little bit of the best of both worlds.
28:10
You get the the angel investor CEO, but you get somebody who's like managing 70 billion in capital. >> Right.
28:15
>> Um, and so, I think the takeaway if you're a founder is generally speaking, at the top firms, the top people are really, really good.
28:26
So, that's So, I think you can be pretty Like if you're if you're deciding between General Catalyst and Lightspeed, like Ravi's like a you know, really smart, amazing.
28:36
You can't really go wrong both the ways.
28:39
Um, there are a few investors who are both well-known and are also terrible. >> Yeah.
28:45
>> But we won't talk about that.
28:46
>> [laughter] >> With that I mean, with that it's just been this wonderful thing on X of finally revealing, you know, the atrocious behavior of these >> Well, there was some bad ones.
28:53
>> Yeah, yeah, there's the YC there's a investor database. >> Yeah.
28:56
>> And it's an anonymous database.
28:56
So it's like it's it's not anonymous to YC, that's to the partners, but it's anonymous to everybody else who's in the YC universe.
29:02
And you'll see scathing reviews on people and they're generally accurate, yeah.
29:06
I I think I think founders are more correct than like holding a vendetta. >> Yeah.
29:11
>> I would say that that that YC kind of internal database on on venture investors is is probably the most important thing for founders in a way that and I wish there was more of that where, you know, we we were very conscious of that always at Giant.
29:22
It's like if you mess with a YC founder, it's going on that database and you'll be blacklisted.
29:28
You're never getting into any YC company again.
29:29
And >> It's like the Uber driver versus a taxi. It's the same thing.
29:33
>> not enough of that I think for for founders.
29:35
>> But I the there's there you know, in the in industry like this you have to be super super careful.
29:40
I mean, we're really talking about like this is real inside baseball.
29:43
Like reputations are a tricky thing because um if I say hey, you know, uh whatever investors really great and they treat one of your friends really poorly, both things can actually be true.
29:57
And then you're now you have a term sheet from this situation where you have an investor who's maybe famous but also is like you know, your friend had a bad experience with them and what's what's the truth?
30:07
And and that's really really that's that's that's difficult that's the fog of war, right? >> To- totally, yes.
30:12
Let's round out this first episode with just a bit of future gazing on physical AI and on applied intuition.
30:20
So what what do you think this is going to do physical AI to change how the listeners to this podcast are going to live their lives?
30:26
How how is this going to fundamentally change things perhaps in ways that people don't quite grasp yet?
30:31
>> I think uh you know, if we were having this conversation 5 years ago, there'd still be a debate on is is self-driving going to happen?
30:37
Is it going to be safe enough?
30:39
And Uh, will the regulators be able to You notice that we didn't ask any of those questions in in you know 30 30 minutes because is we just assume it's going to happen. And that's a big thing.
30:50
Um, for listener you want to extrapolate that on to Okay, so I kind of get the robo taxi thing.
30:55
But what else is going to What about the car that I own?
30:57
What about the construction site that I drive by?
31:02
What about the truck that's on the highway with me?
31:05
Now I go to the airport, why is my luggage moved by humans?
31:07
And then you know, then you get on the plane and so all of everywhere there's a machine, just think about like there's going to be intelligence being put on to it.
31:15
And uh, and I think that's going to have pretty pretty profound impacts.
31:18
And um, as a let's say a corporate executive or as a banker or consultant, I think there's knock-on effects on all those things.
31:27
The mistake that people tend to make though is they expect the same the issue in the physical AI world is diffusion.
31:35
It's like getting you know, getting intelligence on to an airliner is not the same as getting, you know, um, something through a browser.
31:43
Because the whole infrastructure already exists.
31:44
There's Apple Pay and Google Pay.
31:46
There is these operating systems which are quite mature.
31:49
There's the devices that physically are in everybody's pockets and in their homes.
31:54
And so distribution is is very very fast.
31:56
To get intelligence onto a plane or an F-16, that's extremely that's extremely difficult.
32:02
There's lots of gates to to get there.
32:05
But the opposite is also true.
32:05
Once you're in as a company providing the intelligence like a plane applied, then you're really in.
32:10
And because you uh, went through not only the how difficult it is to make models that are efficient and and uh, and and effective and performant, but you also have to, you know, build the relationship with the company and the company has to trust you and and so so that ends up becoming its own moat, frankly speaking.
32:31
>> You just got to talk at Goodwood, the Festival of Speed, which is one of uh, you [laughter] know, conference or event for for car enthusiasts, particularly old car enthusiasts.
32:38
People who love, you know, old cars and love driving. >> Yeah.
32:41
>> Do you think in 10 years, 20 years, anyone will drive at all?
32:43
Will it Will it just be a few tinkerers who love old cars?
32:46
Do you think regular people will be driving?
32:48
>> It will be much different than it is today.
32:50
That's that's that's a shocking thing.
32:52
I mean, in the sense of like uh you know, I'll speak from a personal experience. Like, what do I drive?
32:56
I just got a 1987 Land Cruiser, uh you know, manual diesel.
33:01
In In In Europe, that's like a very passive, but in America, there's You can't even buy those cars.
33:06
The Land Cruisers are not Those versions of Land Cruiser are not sold. So, I had to import it.
33:10
Why am I not driving a a Tesla Plaid, you know, or whatever?
33:12
It's like consumer behavior is not just, you know, Porsche tried to get a get get out of the, you know, manual game.
33:22
GT3s, they stopped making with manuals, and then there was like an uproar. >> Mhm.
33:26
>> It's a worse technology.
33:26
PDK is a better technology. It shifts gears faster. It makes you faster.
33:32
But that's not you know, the re- the reason that It's like, why do people play vinyl records, right? >> Right.
33:37
>> And so, I think it's too simple to reduce everyone's I'll speak consumers desires to what is the fastest, most automated thing.
33:49
>> [clears throat] >> So, that's a broad broad point.
33:49
More specifically, uh or more tactically pragmatically, a car's shelf life is about 15 years, at least in America.
33:58
Half of Americans live on paycheck-to-paycheck or like a limited savings, uh more more accurately.
34:01
And so, when they buy a car, it's a real It's not a luxury purchase.
34:05
It is a functional purchase to live their life.
34:10
And so, you can come with a new product that has more automation.
34:12
Doesn't mean that they're going to sell their Honda Accord because maybe they don't want to sell the Honda Accord.
34:18
Maybe they can't buy the new thing.
34:19
So, you're talking about It takes a decade for just the cars that are bought today, in 2026, to make it out of the cycle.
34:27
And realistically, more like 15 years, 20 years.
34:29
Cars are really really good products right now.
34:31
Extremely cheap for what what you get.
34:33
Uh And so I I I I think but what you but at the same time as somebody who's like lives in the self-driving universe. >> Yeah.
34:43
>> Think when we're here in 10 years from now and you're walking around London, there's no way you won't see many robotaxis, many brands, personally owned, shared systems that are not completely autonomous.
34:55
I I think that will happen actually. It'll be mixed fleets.
34:59
>> And last question for part one, what is the the vision for the impact the player is going to have on the world?
35:05
>> I mean I think we want to put you know, our intelligence on a billion machines.
35:10
And and why do we want to do that?
35:11
Because it makes the world a better place.
35:14
It's so pithy and almost reductive.
35:19
But these are these are dangerous machines.
35:21
So we're going to make the world a safer place.
35:23
I think you don't need to have a long conversation with somebody who's had somebody in a workplace accident or in a car accident or truck, you know, and it's it's horrible.
35:35
It's so bad that you don't think about it.
35:37
Actually cross your mind and your brain just doesn't want to like actually live with the feeling cuz it's so so atrocious.
35:46
That's a pretty that's a pretty good way to live live a life.
35:48
If we can do that, I think that that that's a that's very very positive.
35:52
And we also do believe I think beyond the raw safety aspect of it, I think as efficiency is brought in into people's lives, they live better lives.
36:03
They live just literally happier lives.
36:04
And not only for the the bad stuff, you won't be, you know, injured or maimed, but you just have a happier life.
36:08
And for all the, you know, let's say criticisms we have of modern technology.
36:15
We always talk, you know, that something very like let's say a divisive like a social media.
36:22
Well, you know, the nice thing is you can get a hold of anybody for free around the world instantly.
36:26
Like we forget about that.
36:27
You know, when my family moved from Pakistan or some small village in Pakistan to to America, like we communicated to Pakistan with handwritten letters that would take 30 days to circumnavigate the globe.
36:42
And today it's like WhatsApp is free. >> Yeah.
36:46
>> Like that is a miracle.
36:46
And I think we don't like appreciate that enough because I think it's just And it's not I'm not saying like, you know, everyone should just shut up and eat it and technology is great.
36:56
I'm also not like simplifying that much, but >> Yeah.
37:02
>> we live in a pretty amazing time.
37:02
Like if you have some sort of pain in your body, you can immediately open up a supercomputer [laughter] and say, "These are all the things I'm feeling.
37:15
Should I go to the doctor?"
37:15
That's that's unbelievable.
37:16
You know, for a vast majority of humanity 5 years ago, 10 years ago, that was like not not even comparable.
37:25
Now we have these cheap phones that exist everywhere and those cheap phones can access these really like really significant models.
37:30
So, I think hopefully, you know, near the end of hopefully you can live a long life and at the near the end of life we can look back and say like, "Hey, we made the world a safer place.
37:41
We made it more efficient."
37:41
And And you know, that's that's a that's a great great existence.
37:47
I mean, that's a pretty great way to live.
37:50
>> You mentioned your upbringing.
37:50
It was a masterful broadcast teaser for part two, which is going to be coming out next week.
37:56
And we're going to cuz you've had this incredible life journey, we're going to talk about that.
37:58
We're going to talk about running YC and all of the lessons from YC.
38:03
But but most importantly, I think you've probably got the best career advice for our listeners of all of our guests.
38:07
So, we're going to talk about that next week on part two.
38:09
But thank you so much for joining us for part one. [laughter]