E371: Midas List VC: Why AI Models Will NOT Become Commodities

0:00Hans, you've been on the Midas list several times. You're one of the top VCs in Silicon Valley. And several years ago, you had a chance to invest into open AI and Anthropic. You chose Anthropic. What made you choose to invest in Anthropic at the time? I've been a consumer investor most of my life because the era I've been in, whereas I'm an investor, overlap with the rise of mobile internet, not only here in the US, but also in Asia. And so it's very easy for people who make money from the things that you have done serve you well to continue the same paradigm. So when it comes between OpenAI and Anthropic, both were amazing companies at the time.

0:41This is 2024. people would assume that I would just pick the biggest consumer play at that time in AI and the fastest growing consumer play ever. But with Anthropic, what we saw was something that was to us very interesting. And to me personally, a consumer internet kind of signal that wasn't obvious to people. The API business for Anthropic in 2024 was their smallest business in terms of revenue size. But what's interesting about that business, you can see a lot of developers peeing Claude for codes for things. And Cursor and Lovable and others were going rapidly on top of that. For someone who's more of an aspiring historian, remember the Intel Inside, what Google did with Yahoo's traffic and so forth.

1:21Being embedded in something else and get the benefit of the crowd to keep on peeing you for things. Over time, you can imagine, as I did, that Claude would get, quote unquote, smarter because of different folks, different developers and different verticals asking them for codes. Each one needs to be tailored over time, tailored more towards that vertical. And within enterprises, you can see people paying stuff from different functions, whether it's marketing or product design or coding or security, what have you. And Cloud will understand better and better what people are asking for for the particular function.

1:55So when you have that over a thousand, 10 ,000, a million X, you're more likely to generate a positive feedback loop that will help you to get better at delivering that kind of benefits to the users. And this is what I saw with TikTok. This is what I saw with Airbnb. All the best kind of consumer in the company over the last 10, 15, 20 years has that possible feedback loop that make them build a moat from the wisdom of the crowd and the feedback from the crowd and reacting to what people are asking for. And we saw that with Anthropic, with OpenAI was being doing a bunch of different things. And it was building an empire, which is great, good for them.

2:29But Anthropic was very focused on developer. When we decided to invest in Anthropic, DeepSeek just came out literally a few days earlier. And the token cost on DeepSeek is a lot cheaper. So if the token cost in the US models, closed models are more expensive, you have to take that to generate a lot of value. And what's the best value you can generate in an economy like in the United States? It is building B2B application for enterprise use. In full disclosure, I'm also an investor. My thesis was around B2B versus B2C. But what you're essentially saying is what you saw in Anthropic was that It was using its own recursive improvement in the code to basically improve the system?

3:05Is that your story? And the network effect that was created with all the developers keep on using it. And all the people, you know, 20, 30 years ago used Yahoo because Google embedded inside Yahoo just became smarter and smarter about people were searching for. So Yahoo was doing many different things. But what was most strategic at that point in time it went down to was search. It wasn't obvious at the time. Yahoo was doing so many things, mail, news, and, you know, whatever you can think of today, Yahoo had it back then. But the single most important thing that people back then really wanted was search.

3:32And Google nailed that as part of Yahoo. And Bing tried to do this 10 years later. It just was too late. Even though Bing was great. Amazing engineers. Great, great model. But a great algorithm. But it didn't have the feedback loop and the network effect that Google had as part of Yahoo initially. And then when it spent out, continued to be able to capitalize on that. So those are things that just remind me a lot. This anthropic could have a chance to be hopefully bigger than Yahoo. who know how long it will take, but it's a pro state half. Makes sense. And obviously, I love Anthropic, but I'm always skeptical when people talk about these soft things like values or alignment, all these things.

4:07These are ultimately for-profits. What does it mean when you say that Anthropic essentially had better values? I'll give you another example. Airbnb, March 2020. I was on a panel, another amazing publication with amazing journalists, had four of us on. The other three, the topic of the panel, future of travel in April 2020. So, yeah. Great timing. The other three understandably poo-pooed Airbnb, poo-pooed travel. Like, come on. That was one point in time. At some point, COVID is going to go away. I don't know how long it will take, but it was one point in time. And even back then, is there a chance for Airbnb to still survive?

4:45What could Airbnb do in that time period? Not only how long COVID will last. And when a question was posed to me, I said, you know, staycation could be a thing. I know from first principles, Most consumers, especially American consumers, cannot be cooped up in a house for two, three, four, five months at a time. People need a place to go. Sure, they cannot fly to Paris, to London anymore. But people can drive. We all have cars. Going and driving 50 miles, 100 miles, 200 miles. We'll get to some place where they can find a place to get something different. Now, who in that world can quickly pivot and have new inventory and supply that's in a different place, different location that was popular before?

5:19Only everybody can do that. You cannot just build a hotel like that, that quickly. So they have a chance to do something that other people cannot do and provide for a pent-up user. So when they needed it, who knows if Airbnb can execute, but if Airbnb can pull it off, it will be them. Literally two days later, Airbnb made an announcement, had funding from Silver Lake, in Comfortable Note, and then all the reservations were canceled so that consumers benefited from not being on the hook to pay for these things. And Airbnb used that money to pay the host for the cancellations to keep them in the business.

5:53Then I had to let go, unfortunately, 25 % of the employees at the same time. And I mean, it went out of their way to help come up with a book of all the employees who were unfortunately let go and helped to find a job. Now that people can work remotely in that process. Everything that happened reflect the ethos of Brian Chesky. And nobody in the history of travel had ever done that before. So when we talk about why does values matter? We're talking about dollars and cents and zeros and ones. That's why it matters. And a moment like that builds kind of a in the practical terms, loyalty and trust that you just won't be able to get easily anywhere else, anytime else.

6:29This is why I have several IPOs coming out also in SpaceX and Cerebris. And although I'm excited for them to finally go public, I do have this skepticism on the public markets and the inherent corruption of the long-term value of the company giving the quarterly financials. Do you see that evolving? Is there any way to solve this issue of going public while also thinking long-term? It's a good question. I'm not sure I'm the best person to answer that. You can ask other CIOs about that. I haven't gone through six, seven IPOs already. It's really not easy. The grind of going through every quarter and focus on the question you're going to get every quarter on those earning calls, the pressure is very, very high.

7:11And it takes a lot of time to do each one of them well, which is time taken away from running your business, making all their best. I think the best person I can think of that can manage that and have the sort of the mental fortitude. And this is not a earth shattering answer is Chen Seng Huang. I really remember him 2019 in KOTU's East West Summit where he talks about why GPUs matter and why GPU makes sense for the different kind of verticals that they're in. This is way before AI and before COVID. The way he talks about things and the way he thought about the world, the way he was being willing to make his bed and be lonely for 15 years doing that, that is a complete different world and be able to know that you're right other people are actually going to be wrong and be able to hold that view for 15 years like that when it's kind of lonely no one cares about you it's something that just is very unusual and to see someone like that with similar background as me born in Taiwan grew up outside of the US and then came to the US and had the chance to go to the best schools in the US and really benefit from the American system and to get back as being the largest market cap company in the world as American um naturalized american it is just extremely inspiring and that's a great story for any immigrant who come to this great country uh for for that kind of opportunity but it's a lonely journey and when i look at myself my career with places i've been to a lot of choices are not popular and not understood at the time but you gotta have that mental fortitude to march to a different beat and that's the hardest thing a lot of people buy high sell low chase the latest the thing even i did that as early stage in my career doing whatever's the hardest thing at the time and realize that that just that does not work you got to be able to chart your own path and have afforded to do that.

8:47So, you know, it's very hard to do that in current environment, it takes a different kind of CEO to make that work. Expert calls have always been one of the most powerful ways to build conviction. But today investors are asked to cover more companies move faster and do it with leaner teams. With Alpha Sense AI led expert calls, their TGIS call service team sources experts based on your research criteria, and let's the AI interviewer get to work. The magic is in the AI interviewer purpose built and knowledgeable based information to conduct high-quality context-rich conversations on your behalf, acting as a trusted extension of your team.

9:21Then they take it one step further. Your call transcripts flow natively into your AlphaSense experience and become querible, searchable, and comparable, so your primary insights plug directly into earnings preps, digital work streams, and pitch books with zero tool switching. And with AlphaSense expert call services, the AI-led expert calls are just one option, because we know the importance of a hybrid expert research approach. AI for coverage and efficiency. Humans for complexity and conviction. It's the institutional edge that scales research without scaling headcount. For hedge funds, that means validating thesis assumptions across dozens of experts before earnings instead of a handful.

9:58For private equity, it means faster pre-IOI scans and deeper commercial diligence. For investment banks and asset managers, it means pulling real operator perspectives straight into models and sector positioning without disconnected tools or manual handoffs. All of it lives inside the AlphaSense platform, trusted by 75 % of the world's top hedge funds alongside filings, broker research, news, and more than 240 ,000 expert call transcripts, turning raw conversations into comparable, auditable insight. Take advantage of AlphaSense AI-led expert calls now. The first to see wins, the rest follow. Learn more at alpha-sense.com slash how I invest.

10:38And the only example that I could really think of is Jeff Bezos. The reason Jeff Bezos made it work is he started from the very beginning. Right. He, you know, Mike Maples talked about this, how he picked his LPs for the first couple of funds. He said, you need to find your champions. Right. Some people focus too much on selling and not enough on picking who your right partners are. And I think Bezos did it from the beginning. He hired the people around him from the beginning. The antithesis of this was actually Facebook. if you remember I got to sit down with Roger McNamee from Elevation he sat me down in the chair where convinced Mark Zuckerberg not to take the billion dollar deal everybody around Mark Zuckerberg wanted him to take the deal and he fired most of those people I know I saw one more I'm not going to name names it's as much as a financial thing it's a cultural thing and I think there's also well there's also a mental aspect to it what does that mean why is it that I'm also a first generation immigrant why is it that immigrants have that much power and have that much success in Silicon Valley despite all the things going against them.

11:42I think one of them is being different and being an other cannot be taught. Right. You have to feel it. And a lot of people you grew up in an environment you became comfortable with to leave that place of comfort where everything's familiar to you to go to something completely different and you can immediately feel you're an outsider, you're the other and try to see through that and analyze that and figure out how to fit in but how to also be different by being able to step out. That training, doing that, and that's why I would like the Gladwell's books, 10 ,000 hours. You have that training when you were 13, you were 10, to all the way when you have to work.

12:14You have 10 ,000 hours of thinking that you're different, but how do you turn on your advantage? That is invaluable to work in an environment that's tough, that's unfamiliar, where your view is minority point of view. 50 % was the last statistic that I saw in terms of first or second generation immigrants venture funded. And then you add to that the tealism about human beings are so mimetic and so into pleasing others and making other people feel good that you almost have to be neurodivergent in order to go against the grain. So now you have 50%. Now you have, I'm not going to ask you for a percentage of neurodivergent founders, but some percentage of that.

12:47And then what's left, other obviously modes of thinking and other types of diversity as well outside of just being an immigrant. Then the question becomes, can a regular person start a$100 billion company with no trauma, with no background, with no otherness in their past? And I'm not sure that that's possible. I tried to think as you mentioned this. Any normal founders in your portfolio? Who have, I think a normal, quote on, normal founder can get to that half a billion, 200, a billion. And branching is also different. That's the same. We put on your bed, you know, you're a home for a stranger.

13:22That does not do that. And I then have the wisdom to know that's right. When the first time I heard about it, going to YC, I'm like, that's the stupidest idea ever. But you see how the world changes to adapt to that. Okay, okay, I was wrong before, you got to make changes. When I'm still investing in him later. So, you know, thank God we are, you got to have the, the thick face to say, yeah, I was wrong and we got it. We got to evolve and be able to iterate that. That's also extremely important. And the corollary to that is that yes, to be different, give you an edge, but also are you intellectually honest?

13:52It's not about pride. It's about figuring out, look at everything from first principles. What did I miss before? How does that change the way I think about things? Let's change the original assumptions to see what, what does the output look like the second time. I had a fascinating conversation with Eric Scott, who is a founder's fund. Yeah. And he explained, and within founder's fund, they have this unique ability to look at every round from complete first principles. In other words, there are companies where you were wrong to pass on the seed. You were right. Support for today's episode comes from Square, the all-in-one way for business owners to take payments, book appointments, manage staff, and keep everything running in one place.

14:27Whether you're selling lattes, cutting hair, running a boutique, or managing a service business, Square helps you run your business without running yourself into the ground. I was actually thinking about this the other day when I stopped by a local cafe here. They use Square, and everything just works. Checkout is fast, receipts are instant, and sometimes I even get loyalty rewards automatically. There's something about businesses that use Square. They just feel more put together. The experience is smoother for them, and it's smoother for me as a customer. Square makes it easy to sell wherever your customers are, in-store, online, on your phone, or even at pop-ups, and everything stays synced in real time.

15:04You can track sales, manage inventory, book appointments, and see reports instantly whether you're in your shop or on the go. And when you make a sale, you don't have to wait days to get paid. Square gives you fast access to your earnings through Square checking. They also have built-in tools like loyalty and marketing, so your best customers keep coming back. And right now you can get up to$200 off Square hardware when you sign up at square.com slash go slash how I invest. At Square, you get all the tools to run your business with none of the contracts nor complexity. Run your business smarter with Square.

15:35Get started today. Support for today's episode comes from Square, the all-in-one way for business owners to take payments, book appointments, manage staff, and keep everything running in one place. Whether you're selling lattes, cutting hair, running a boutique, or managing a service business, Square helps you run your business without running yourself into the ground. I was actually thinking about this the other day when I stopped by a local cafe here. They use Square, and everything just works. Checkout is fast, receipts are instant, and sometimes I even get loyalty rewards automatically. There's something about businesses that use Square.

16:06They just feel more put together. The experience is smoother for them, and it's smoother for me as a customer. Square makes it easy to sell wherever your customers are, in-store, online, on your phone, or even at pop-ups, and everything stays synced in real time. You can track sales, manage inventory, book appointments, and see reports instantly whether you're in your shop or on the go. And when you make a sale, you don't have to wait days to get paid. Square gives you fast access to your earnings through Square checking. They also have built-in tools like loyalty and marketing, so your best customers keep coming back.

16:38And right now you can get up to$200 off Square hardware when you sign up at square.com slash go slash how I invest. At Square, you get all the tools to run your business with none of the contracts nor complexity. Run your business smarter with Square. Get started today. Support for today's episode comes from Square, the all-in-one way for business owners to take payments, book appointments, manage staff, and keep everything running in one place. Whether you're selling lattes, cutting hair, running a boutique, or managing a service business, Square helps you run your business without running yourself into the ground.

17:09I was actually thinking about this the other day when I stopped by a local cafe here. They use Square, and everything just works. Checkout is fast, receipts are instant, and sometimes I even get loyalty rewards automatically. There's something about businesses that use Square. They just feel more put together. The experience is smoother for them, and it's smoother for me as a customer. Square makes it easy to sell wherever your customers are, in-store, online, on your phone, or even at pop-ups, and everything stays synced in real time. You can track sales, manage inventory, book appointments, and see reports instantly whether you're in your shop or on the go.

17:44And when you make a sale, you don't have to wait days to get paid. Square gives you fast access to your earnings through Square checking. They also have built-in tools like loyalty and marketing, so your best customers keep coming back. and right now you can get up to$200 off Square hardware when you sign up at square.com slash go slash how I invest. At Square you get all the tools to run your business with none of the contracts nor complexity. Run your business smarter with Square. Get started today. To pass on the A you were right to pass on the B and then you were wrong again. And just the ability to constantly reevaluate your priors and think not in a deterministic manner but in a probabilistic manners.

18:21Most VCs can't grasp their head. Right. I mean, the common thing these days is power a lot. That's not rocket science. We're in a business of hitting home runs, ideally, but base loaded. So it's a loading position that matters. But how do you reconcile that with being stuff early in a very healthy ownership level? And a lot of times you have healthy ownership of stuff that's more second tier, not the best tier. And so we have double down the best tier companies versus find the next second tier companies but you own a lot and tell people you're a great early state investor. How do you reconcile those two?

18:54It is not the easiest thing. And for me, I'm not ashamed to admit it, I don't have the highest ownership in the companies that you mentioned are great names, but I feel like when they're early enough, we got to be in them for the right reasons. And I wish I could put more money in them, but we have the partnership to talk about all these kind of issues. And in the previous era of thinking, having own more in a company is better. These days, because some of the companies that can be so much bigger, you almost have to see that, okay, these are the best companies and being in them is important and having to pile more money in, you should because the upside from the point of second entry, third entry is still like 10X.

19:30So you got to do that. And that is not easy for most people to admit that that's what you should be doing. We're over 300 episodes into this podcast. And early on, I used to ask this question. People would answer the same way. And then after a while, I still didn't understand it, but I stopped asking it. The question is, people always talk about your fund size is your strategy. And they talk a lot about ownership but why does it not just come down to valuation? I understand you can't just invest$25 ,000 and hope it 100Xs, but it seemed like everybody was just using different terminology to back into being valuation sensitive at the early stage.

20:05Right. Why do people talk about it in terms of ownership and fund construction? Why do they overcomplicate what I think is fairly straightforward? Because it's so hard to have winners. And when you have winners, at least at that point in time, all LPs ask, well, what is the ownership level? So out of 10 bets, 15 bets, or 20 bets, you got one or two right, and you own 20 % each, and ends up being not 100x, but 500x or 1 ,000x as an early, very early seed investor, then the math works out. But isn't that backwards looking in that if you just had 30 companies instead of 15, and you own 10 % instead of 20%, isn't that the same thing outside of just trying to recreate this analogy where you would know the one company that you should own more?

20:42Right. You could do that, but the challenge is that, are you sure your 15 really captured enough diversity to give you a chance to hit the one that's really big? 30 seems to give you a higher chance of at least hitting one of them big. Yeah. So if you want to hit 30, then you also want to make sure you own enough, each one of the 30, which means that enterprise has to be lower to give you a chance to do that if the fund size is fixed. So I can see the argument, and I can say I'm a great seed investor because that is not ideal for living, even though like Xiaomi and Red Note and some of the stuff I got in, it was below$100 million valuation.

21:13And they ends up making a lot of money if it ends up being big. So I can see the argument of what you have shared, but it's not easy to find stuff that really works until a little bit later. And having the flexibility to do that is very, very important in my mind. And also knowing where to hunt, where the world's heading, therefore you got to spend more time doing that is a luxury that a lot of people don't want to try to bet on because they don't know what could be big, what could be interesting. So it ends up spreading the bet. Everyone's trying to figure out who's going to win in the world of AI.

21:40In terms of AI native apps, how do you build for this constantly evolving LLMs? It's very, very difficult. And I think a lot of founders out there who are AI native companies constantly think about that. And that's probably the number one reason, as you can imagine, to keep them up at night. One, you want to be on the token path. If the models are improving, your product builds on top of that. As models advance, your product also gets a lot better and generates a lot of value, like a cursor, like livable, and so forth for coding. But the worry is that if cloud and cloud code and cloud code work, it's so much better than what is the room that is left available for you to do.

22:14And for that value, one way in theory is that. The adoption at a user level vary, the people more technical, the people a lot less technical. So if you simplify the UI and reach more mass users, mass market users, there's a chance you can stay ahead of the UI of a more, something more technical from cloud possible. Another way is that you actually own a system of records for data. What is transactions, content, what have you. Another way is that you actually make the final buy and sell decisions, risking the balance sheet of individual household or enterprise. And therefore, LM don't want to touch that kind of balance sheet risk.

22:54So these are ways I can think of. And then lastly, have all that domain knowledge, something that's more regulated, all the compliance factors and so forth. And LM don't want to spend time doing that. Actually, I think about it. There's not one more. Physical AI is possible. Because in physical AI, one can argue, it's not easy to have one model that rules at all, one real that rules at all. Therefore, in different situations, can you exist where it's harder for LM to go after you? These are the things I can think of today. But, you know, a year from now, two years, a month, a month, a month, who knows?

23:19And that's the best guess we can come with at the current moment in time. Physical AI blows my mind. What's exciting to you about physical AI? Physical AI, it definitely lacks behind software AI. and in the U.S., when we still have the best engineers in the world who want to come here to build amazing companies and 70 % of U.S. economy is tied to services and it's not, it doesn't make a rocket scientist to know that, you know, serving the developers build B2B enterprise applications is the most value generating activity in a service oriented economy. But the rest of the world is not like that.

23:49And the rest of the world is over 90 % of GDP outside of U.S. is tied to physical industries. And therefore, if you also think that the second rule is not like they have one model that rules at all, um then for each different kind of physical industry there may be different models that can emerge the work center for that industry at least there's a shorter path of commercialization that's less capital intensive give there you know god blessed uh uh lacune for coming up with something interesting and big and he can still do that or uh world models or or or pi but short outside of that more likely is going to be around the world will be smaller models for specific things and that give more people a chance to win and that that's what to me is exciting for venture as well.