Two of the more candid voices in early-stage AI sat down last week to walk through their AI startup investing strategy, and neither of them made it sound remotely straightforward.
Carter Reum, co-founder of M13, and Chang Xu, a partner at Basis Set Ventures, appeared at TechCrunch’s StrictlyVC evening in El Segundo, Los Angeles. M13 manages $2.5 billion in assets and has backed 17 unicorns at seed or Series A. Basis Set launched in 2017 as one of the first early-stage funds focused exclusively on AI and recently closed its fourth fund at $250 million, up from $185 million for the prior fund, according to the Wall Street Journal, bringing total assets under management to nearly $1 billion.
Pricing Deals When Revenue Growth Defies Old Models
The central tension both investors keep running into is valuation. When a company can move from zero to meaningful ARR in months, standard multiples feel both justified and dangerous at the same time.
Xu pointed to OpenArt as the clearest illustration of what genuine velocity looks like. The generative-image platform went from $1 million to $10 million ARR in its first year, then from $10 million to $70 million in year two, cash-flow positive for most of that stretch with just 20 people. OpenArt has since raised a $30 million Series A led by Canaan and now counts over 8 million monthly active users, according to Basis Set Ventures.
‘The bar for what is good growth has totally changed,’ Xu said. ‘When you have this possibility of compounding accelerant growth, the valuations don’t seem so crazy because you price that into the terminal value. On the other hand, if you price every single deal to that math, there’s no way that will work out well for a portfolio.’
Reum applies what he calls cocktail-napkin math: asking how big the winners were in the last cycle, whether the market is structurally growing, and whether buyers will pay meaningfully more for software this time around. He described walking away from an AI software deal targeting brands precisely because that arithmetic wouldn’t close.
The Frameworks Behind Their AI Startup Investing Strategy
Xu’s most transferable framework for any AI startup investing strategy is the distinction between velocity markets and depth markets. In a velocity market, fast followers move faster than ever, so execution speed is the only real moat. In a depth market, hard things remain hard regardless of how much compute you throw at them.
She illustrated the depth-market idea with a portfolio company using transgenic chickens to manufacture complex proteins more cheaply than conventional bioreactors allow. ‘Chickens still take this long to hatch…,’ she said. The biological clock is the moat, and no large language model is about to change that.
Reum’s parallel concept is friction as a moat. M13 targets regulated industries precisely because hyperscalers are reluctant to enter them at smaller scale. The firm recorded an exit just shy of $1 billion from a company that brought AI to 911 call centres. Healthcare is another deliberate target. ‘The hyperscalers might go there eventually, but as a few-billion-dollar outcome, they’re not going there anytime soon,’ he said.
Both investors agreed the competitive structure of this cycle is unlike any prior one. In every previous tech wave, innovators competed with other innovators. Now they compete with each other and with the ten largest, best-capitalised technology companies ever assembled. ‘For the first time in history, the incumbents actually do have the advantage: the tech, the capital, the data, the talent,’ Reum said.
Xu’s framework for staying defensible is a constant focus on technical differentiation, revisited every quarter, sometimes every month. She draws a line between investing below the AI, meaning infrastructure being rebuilt for agents rather than humans, and above the AI, meaning application layers with long-term moats. A new version-control system built specifically for AI agents, she said, would have seemed absurd twelve months ago. She can now count on two hands the strong teams pursuing exactly that.
M13’s fourth fund, M13 Ventures IV, filed a Form D with the SEC listing a total offering amount of $400 million, according to a Form D filing via 13f.info. That sits within M13’s overall $2.5 billion in assets under management across funds.
What the SpaceX IPO Could Mean for Los Angeles
The conversation ended on geography. The anticipated SpaceX IPO is expected to distribute wealth more widely across the Los Angeles workforce than previous tech listings, which concentrated returns among institutional investors and a narrow group of VCs.
Reum’s view is that every major liquidity event seeds the next generation of founders. The previous LA cycle produced Riot Games, Tinder, and Snap. ‘This is a different order of magnitude,’ he said. Three years ago, observers were writing off San Francisco. It turned out to be less dead than advertised, and Reum argued the same would be true for anyone inclined to dismiss LA.
Xu framed the city’s coming moment around a single word: taste. As models get better at automating engineering work, the next frontier in AI is not more compute but cultural resonance: films, videos, content that connects emotionally with specific audiences. San Francisco dominates on technical talent. LA dominates on something the models cannot yet replicate.
Reum’s parting image was a stone skipping across water. The heavier the rock and the faster it’s thrown, the longer the ripples. The first wave is crowded and obvious. The second and third are where the returns concentrate, and where far fewer investors are paying attention. His bet is that the most interesting companies of this cycle haven’t been founded yet.
