Investing in AI’s speed: pricing, moats, liquidity

How to – In El Segundo, two early-stage AI investors argued the market is both unprecedented and punishing: growth metrics can look unreal, but sustainability is still a gamble. Carter Reum and Chang Xu mapped how they price deals, where they look for real defensibilit

The room was sunlit in El Segundo, but the conversation didn’t feel calm. Carter Reum and Chang Xu—both investors who specialize in AI—kept circling the same problem: venture capital is trying to price companies in a market that changes so fast it can make yesterday’s “obvious” assumptions feel outdated before the pitch deck cools.

Reum, co-founder of M13, started with the basic math of why the speed feels different now. M13 is an early-stage firm with $2.5 billion in assets under management. and he said it has backed seed or Series A investments in 17 unicorns. On stage, he framed the moment as familiar in shape but harsher in pace. “We’ve seen this before. ” he said. pointing to cloud. the iPhone. and the car in the 1920s—periods where people worried about jobs. and then life moved on. The difference now, he argued, is the matchup.

In this cycle, he said, innovators aren’t just competing with other innovators. They’re also up against the largest. most well-funded innovators on the planet—and the ten largest tech companies on the planet. “For the first time in history. ” he said. “the incumbents actually do have the advantage—the tech. the capital. the data. the talent.” His warning wasn’t that growth won’t happen. It was that it can be followed by sudden collapse, making the investment window feel narrower.

Chang Xu, a partner at Basis Set Ventures, pushed on the same tension from the other direction. Basis Set Ventures launched in 2017 as one of the first early-stage funds focused exclusively on AI. and Xu said it is now investing out of its fourth fund. with nearly $1 billion in assets under management. He described the “bubble” question as a paradox. “There’s both a bubble and not a bubble,” he said, explaining that the growth curve is unprecedented.

Xu pointed to OpenAI’s ChatGPT as an example of how fast the revenue story can move. saying it went from one to $40 billion in six months in terms of revenue. He also talked about a portfolio company. Open Art. that he said went from $1 million to $10 million ARR in year one. then from $10 million to $70 million in year two. He added that it was cash-flow positive most of that time with just 20 people.

That’s the reason. Xu argued. that valuations can start to look less crazy—if you believe in compounding growth strong enough to reshape terminal value. But he also warned that when you price every deal to that same math. “there’s no way that will work out well for a portfolio.” In other words: the market’s best stories look like miracles. and the average ones still have to earn their way to those numbers.

The pricing problem came next—what do you do when startups generate revenue faster than ever but the durability isn’t clear yet?. Reum said they run what he called “cocktail napkin math,” using a real example from their recent work. He described looking at an AI software business for brands and asking how big the winners were in the prior cycle—whether there will be more brands in the world. and whether brands would pay double or triple for software in this cycle. In the end. he said they didn’t make the investment because they couldn’t get the math to check out.

Xu’s answer to the same question leaned less on top-line pricing and more on defensibility. He said they stay very close to what he called defensible technical differentiation because the frontier changes “every quarter. maybe every month. sometimes every week.” His framework was an investing line drawn across and around AI: investing below the AI and above the AI.

Below the AI. he described infrastructure that is being rebuilt—databases. version control. and deployment tools—because it was built for humans. Now he said agents are using that infrastructure, and agents require fundamentally different things. “Last year I would never have thought you’d need a new GitHub. ” he said. adding that this year he could count on two hands how many really strong teams are going after being “the GitHub for agents.”.

Above the AI, when the market gets crowded, Xu returned to the same hard question: what is defensible and what has long-term differentiation?

That naturally led to a question founders hear in one form or another every day in 2026: how do you build something that won’t get blown apart by OpenAI. Anthropic. or Google?. Reum said he tries to work out where companies are going first and where they’re going last. He argued it was obvious hyperscalers would move into marketing and other obvious places. His thesis instead focuses on “friction as a moat.” He said they love regulated industries and cited what he called a just-shy-of-a-billion-dollar exit in a company disrupting 911 call centers with AI.

He said hyperscalers might go there eventually, but not anytime soon, and he tied the timing to regulation. He added that healthcare is a different story—the major players will eventually get there, but the regulation slows them down.

What kept everyone on edge, Reum said, is how quickly the dynamics can flip. “It can change on a dime,” he said. In his view, the old way of watching the rearview mirror doesn’t apply anymore. So he tells founders they need a “microscope” in one eye and a “telescope” in the other: the microscope for the day-to-day execution—what must be done this week—and the telescope for what’s shifting in the wider world. He framed it as a constant board change, where founders need to play like both domino players and chess players.

Xu gave the same idea a different label. His framework, he said, is whether you’re in a depth market or a velocity market. In velocity markets, fast followers can move even faster—it becomes a race defined by speed of execution. In depth markets, hard things stay hard. He pointed to a portfolio company using transgenic chickens as an alternative to manufacturing drugs. saying it’s expensive to manufacture complex proteins and that it’s cheaper. apparently. if chickens do the work. He joked that chickens still take a long time to hatch “for today.”.

When asked whether founders are bringing genuinely novel ideas or mostly repackaged versions of older companies. Xu said the answer is both. He named the consensus categories—agents applied to finance and agents applied to healthcare—where he said many strong founders are going after and where many are likely to win. But he reserved his strongest language for ideas that don’t initially look like businesses.

He turned again to Open Art. He said that when Basis Set Ventures backed them, shortly after, Dall-E came out and Stable Diffusion came out. He described Open Art’s discovery page of prompts you could type to get certain types of generative images. then asked how that could even be a business—at first. But, he said, Open Art went from $1 million to $70 million in two years and kept accelerating. He credited the company’s timing and iteration. saying that if they’d started a year later they would have missed the window.

The investors also talked about how venture stories repeatedly reinvent themselves. Xu referenced how. years earlier. people said it was a bad idea to invest in anything selling to Hollywood—then later they did deals in creative AI and generative AI. which led to companies doing incredibly well: generative images first. then video. then world models. He said that world ended up bigger than anyone expected when looking back at the prior generation of software selling to Hollywood. He pointed to Cursor. which everyone said was “just an AI wrapper. ” and called its reported $60 billion exit a turning point.

Reum added a reminder about who benefits. He described how when his husband was doing his PhD at MIT, his pay was barely above the poverty line, and now researchers are who everyone follows on Twitter.

Reum then returned to the rhythm of cycles: the first wave is the obvious one. Even in a cycle that feels steep and fast, he said, that initial period brings more competition and crowded markets. The second and third ripples are where the opportunities get more interesting. To explain. he used a physical metaphor: when you throw a heavy rock hard enough to skip across water. the heavier the rock and the faster you throw it. the longer the ripples. He said he gets excited about two. three. and four years from now. when business models and companies may emerge that nobody can imagine today.

In that later period, he argued, the bets are harder for venture capital—but fewer people chase them, valuations are more reasonable, and returns tend to be better.

Then the conversation shifted west—because this was Los Angeles, and because liquidity has its own gravity.

Reum addressed the coming SpaceX IPO and what it could mean for people in L.A., especially employees. He said that when Anthropic and OpenAI “eventually” IPO. it would likely involve a wide spread of money through VCs and institutional investors. But he argued that SpaceX is different: “Never has this much money come back and been so widely spread out as what’s going to happen with SpaceX.” He used a blunt. practical instruction for anyone in the room: if someone has a house to sell. a boat. a plane—take advantage of the ride. His deeper point was that major liquidity events create a second wave.

He cited the previous L.A. cycle’s outputs—Riot Games, Tinder, Snap—and said this one could be a different order of magnitude.

Xu followed the same theme with a different emphasis. He said San Francisco had been declared “dead” three years ago. and it turned out to be “a little less dead than people expected.” He argued the same might be true of L.A. “There are too many smart people here,” he said, both technically and in areas like brand, content, creators, and influence. He described the first wave as a technical one. with technical talent concentrated elsewhere. then asked what comes after technical waves—new business models. creative thinking. and a deeper understanding of culture. He argued that the next wave has a high likelihood of being centered in L.A.

Xu ended by drawing the next frontier line in a way that matched the theme of speed and taste. He said the next frontier in AI isn’t more compute. It’s “taste”—making films, making videos, making things that resonate emotionally and connect with specific cultures. He pointed out that San Francisco has extraordinary technical talent. and models are getting better at automating and accelerating what that talent does. But he said L.A. has “taste in spades.”.

In the end, the stakes weren’t theoretical. They were about timing. About which kind of advantage survives the next turn in the market—technical differentiation. friction in regulated spaces. or an ecosystem that turns money into companies. In a cycle where growth can look instantaneous and reality can change “on a dime. ” that’s the part investors can’t afford to get wrong.

AI investing venture capital Carter Reum M13 Chang Xu Basis Set Ventures OpenArt ChatGPT pricing deals hyperscalers friction moat 911 call centers SpaceX IPO Los Angeles tech ecosystem