TechCrunch Strictly VC Night in Los Angeles late last week brought together two of the most sincere investors working in AI right now. Carter Reum is co-founder of M13an early-stage company with $2.5 billion in assets under management that has been a seed or Series A investor in 17 unicorns, he says. Chang Xu is a partner at Base Set Companieswhich launched in 2017 as one of the first early-stage funds focused exclusively on AI and is now investing from its fourth fund, with nearly $1 billion in assets under management.
On stage, in a sun filled room in El SegundoThe two were as entertaining as they were insightful, covering how to price deals in a market that has never moved so fast, how to find companies that won’t get swept away by hyperscalers, and what the SpaceX IPO is about to do in Los Angeles. The conversation has been condensed and edited for clarity.
Is there an AI infrastructure bubble?
Chang Xu: There is both a bubble and not a bubble. It is not a bubble because we have never seen this type of growth curve before. ChatGPT goes from $1 billion to $40 billion in six months in terms of revenue; That’s unprecedented growth on that scale. We have a portfolio company, Open Art, that went from $1 million to $10 million ARR in the first year, and from $10 million to $70 million in the second year, (and was) cash flow positive most of that time with only 20 people. The bar for what good growth is has totally changed. When you have this possibility of capitalizing on accelerated growth, valuations don’t seem so crazy because that is included in the terminal value. On the other hand, if you price every trade based on those calculations, there’s no way that’s going to work well for a portfolio. So it’s a paradoxical moment.
Carter Reum: I always laugh because we pretend this is the first time in venture capital, but we’ve seen this before: with the cloud, with the iPhone, with the automobile in the 1920s, when people were afraid of losing their jobs, and they did, and life went on. This is steeper and faster, but the same dynamic. What’s different about this cycle is that in previous cycles there were innovators competing with other innovators: Zuck versus Evan, Travis versus John Zimmer. In this cycle there are innovators competing with innovators, competing with the biggest and most well-funded innovators the planet has ever seen. and competing with the ten largest technology companies on the planet. And I would say that for the first time in history, the incumbents really have the advantage: the technology, the capital, the data, the talent. Therefore, as soon as some of these companies grow, they may fall. In fact, I find it more difficult to invest in a market like this. But if you do it right, you look like a genius.
How do you price deals when startups are generating revenue faster than ever but it’s unclear how sustainable they are?
Guilty: We always do the cocktail napkin math. The other day we were analyzing a business: artificial intelligence software for brands. I asked: how big were the winners from the last cycle? Will there be more brands in the world? Are you willing to pay double or triple for software this cycle? We ended up not making the investment because we couldn’t do the math verification.
Xu: We stay very, very close to what the defensible technical differentiation is, because that boundary changes every quarter, maybe every month, sometimes every week. The framework we think about is investing below and above AI. Underneath AI, there is all this infrastructure that is being rethought (databases, version control, deployment tools) because they were all built for humans. Now you have agents using all this infrastructure and agents require fundamentally different things. Last year I would have never thought you would need a new GitHub. This year I can count on two hands how many really strong teams are going after being the GitHub of agents. Above AI, when things get crowded, we always come back to: what is defensible and what has long-term differentiation?
As do Do you invest in companies that won’t be destroyed by OpenAI, Anthropic or Google?
Guilty: We always try to think about where they are going first and where they are going last. It was obvious that they would go after the obvious marketing and places. So we have a thesis about friction as a moat: we love regulated industries. We had an exit of just shy of a billion dollars in a company disrupting 911 call centers with AI. Hyperscalers might get there eventually, but as a result of a few billion dollars, they won’t do so anytime soon. Healthcare: They will go there, but there are a lot of regulations holding them back.
What keeps us all up at night is that it can change in the blink of an eye. You used to see them coming in the rearview mirror. I tell every founder: you need a microscope in one eye and a telescope in the other. The microscope is for everyday life: what I have to do this week, execute. But you better have your telescope outside, because the world is changing very quickly. You have to be a domino and chess player, because your board is constantly changing.
Xu: The framework we use is: is this a depth market or a speed market? In speed markets, fast followers are faster than ever; It’s all about speed of execution. In deep markets, hard things are still hard. In fact, we have a company in our portfolio that uses genetically modified chickens as an alternative to manufacturing drugs, because it is very expensive to manufacture complex proteins. Apparently it’s cheaper if you have chickens, go for it. Chickens still take so long to hatch, today (laughs). Those are deep markets and we invest accordingly.
Chickens notwithstanding, are you seeing any genuinely novel ideas at the moment, or mostly new versions of old companies?
Xu: Both. In the consensus categories (finance agents, healthcare agents), you see a lot of really strong founders chasing them, and a lot of them are going to win. But the most interesting ideas are the ones where you think, ‘Hey, I don’t know if that can even be a business.’ open artwhen we first endorsed them; shortly after, Dall-E came out, Stable Diffusion came out, they started a discovery page for prompts you could write to get certain types of generative images. How is that a business? Absolutely no idea. They went from $1 million to $70 million in two years and have been accelerating ever since. There is so much depth to that market that we simply couldn’t tell from the outside. But from the beginning these were young founders experimenting on the cusp of something they found exciting and kept iterating until they found a business. If they had started a year later, they would have missed the opportunity.
The history of VC is that it is constantly a story of bad ideas becoming good again. Four or five years ago you would have said that it is a bad idea to invest in anything that sells to Hollywood. Then we made a bunch of deals in creative AI, generative AI, which led to the current wave of companies doing incredibly well: first generative images, then videos, and now world models. That world has been much larger than we might have estimated by looking at the previous generation of software sold to Hollywood. And then you have Cursor, which everyone said was just a container for AI. An exit of 60 billion dollars. And the researchers: when my husband was doing his PhD at MIT, his salary was barely above the poverty line. Now researchers are who everyone follows on Twitter.
Guilty: I think we’re still in the early innings. The first wave of any technological cycle, even one this steep and rapid, is usually the most obvious: more competition, more competition. The second and third waves are where it gets interesting. Think about when you were a child: if you take a heavy stone and throw it as hard as you can and make it skip across the water, the heavier the stone and the faster you throw it, the longer the ripples will be. That’s what we’re going to have here. I am excited in two, three, four years, because there will be business models and companies that we cannot imagine today. As venture capital, those second and third bets are the hardest to get right, but if you do, fewer people think about it, you pay more reasonable valuations, and the return on investment tends to be much better.
The SpaceX IPO will put a lot of money into the hands of people who live here in Los Angeles, especially employees. What does that mean for this ecosystem?
Guilty: When Anthropic and OpenAI finally go public, they will be a group of venture capitalists and institutional investors. Never has so much money been recovered and distributed as widely as what will happen with SpaceX. If anyone (in this room) has a house to sell, a boat, a plane, definitely take advantage of that trip. But most importantly, every major liquidity event generates a second wave. The previous Los Angeles cycle produced things like Riot Games, Tinder, Snap. This is a different order of magnitude.
Three years ago everyone said that Saint Francis was dead. Turns out he’s a little less dead than people expected. I think the same would be true for anyone who dismisses Los Angeles. There are too many smart people here, technically, but also people who understand brand, content, creators and influence. This first wave is a technical wave and the technical talent is concentrated elsewhere. But what comes after the technical waves? New business models, creative thinking, understanding of culture. That will be the next wave and I think it will most likely be centered in Los Angeles.
Xu: The interesting thing is that the next frontier in AI is not more computing: it is taste. It’s about making films, making videos, making things that resonate emotionally, making things that connect with specific cultures. San Francisco has extraordinary technical talent, and that’s exactly what models are becoming very good at automating and accelerating. Los Angeles has a lot of taste.
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
