The inevitable has happened: AI has come for YC.
More than 65% of Y Combinator’s summer cohort, totaling 134 startups, are building tools around AI, bringing automation to healthcare claims, customer service automation, sales outreaches, coding and game design.
Investors fear missing a train that is leaving the station, leading to intense competition and vast amounts of capital for unproven startups. From Q1 to Q2, seed-stage deal value for generative AI startups nearly doubled, and deal count increased by more than a third.
“Either you decide to not play this game in AI or you decide, ‘OK, this is the new normal,’ ” said Francesco Mosconi, senior venture partner at Pioneer Fund, which is one of the most active investors in generative AI startups, according to PitchBook data.
A part of that game is acting quickly as rounds fill up quickly. In extreme cases, rounds have been filled the same day a founder decides to raise—cutting out investors who aren’t connected, Mosconi said.
“It feels really bad, especially if you were there first,” said Mosconi, who has personally experienced this kind of close miss.
‘It’s just madness’
“The excitement is only growing,” said Rajiv Bhat, a senior partner at Pioneer Fund who also operates within the AI fund. Bhat said startups are popping up equally across focus areas like healthcare, AI core and horizontal platforms.
The sheer number of startups has made it difficult to keep up. Bhat said the fast-changing nature of advancements in technology, as well as actions by incumbents, adds a difficult layer to seed-stage investing. Bhat pointed to startups focused on the enterprise sector that were deeply affected when OpenAI announced its ChatGPT Enterprise offering.
“Something that seemed great three months ago suddenly looks so dated,” Bhat said. “It’s just madness.”
Big tech giants like Google, Microsoft and Salesforce loom large in the generative AI space, according to Rashmi Gopinath, a general partner at B Capital. She said one of the hardest parts of seed-stage investing in generative AI is assessing a startup’s defensibility.
“Does the startup have what it takes to build a moat for themselves for three to five years? Can it be defensible? Beyond that it could be a great acquisition target,” Gopinath said.
The valuation question
Another key challenge Gopinath pointed out is sky-high valuations that haven’t come down from 2021 levels. She pointed to the levels of excitement for generative AI fueling hundred-million-dollar valuations and the risk that comes with that.
“How do we properly underwrite these companies based on where they’re at in their lifecycle? And how do we not overvalue these companies that puts them at risk for future financing down the road?”
While Mosconi believes valuations are fine, given the enthusiasm for the space, he cautioned that overcrowding in generative AI could become a problem—and that he was already witnessing it with large language model training and infrastructure startups.
“If there’s a gold rush, people say, ‘Hey, build picks and shovels and you’ll be successful no matter what,’ ” Mosconi said. “I think too many founders are starting to pick up on that analogy.”
Featured image by Julia Midkiff/PitchBook News
