'Alien Intelligence': poolside's Founders On The AI Model Race And Where VCs Are Misleading Startups
Jason Warner and Eiso Kant think they can still win the AI model race at poolside. They tell Upstarts how -- and a bunch of other hot takes, from Meta to bad VC advice.
The Upshot
About 45 minutes into my second interview with Jason Warner and Eiso Kant, the founders of AI startup poolside, the conversation turns naturally to whether AI models – which they believe will gobble up so much – are coming for Upstarts, too.
“You’re safer than you think,” Kant, poolside’s CTO, soothes. “Your data moat is asymmetry of information. You’re on this call with us, getting asymmetric info. When you get a revenue number, only you have it. A journalistic scoop is something where human contact is critical, because as humans, we’re willing to share those things.”
Reassuring words, that surely poolside would back up by sharing its own latest sales numbers? The otherwise outspoken Warner and Kant won’t go that far. All their frontier AI lab will say is that it works with companies in defense, financial services, industrial manufacturing, software and the public sector – no numbers, no naming names.
But Warner, the former CTO of GitHub, and Kant, the former founder of multiple startups, have plenty of other hot takes on the AI market – and what it will take for startups to win – that are of interest to anyone building, buying, or competing with AI tools.
At a time when the biggest AI players are dominating headlines – just this week we have OpenAI and Microsoft reportedly locked in high-stakes negotiations; Mark Zuckerberg and Meta shelling out billions for top AI talent; Elon Musk’s xAI fundraising; and Nvidia’s and Anthropic’s CEOs trading barbs – poolside is a dark horse in the race to reach artificial general intelligence, or AGI.
The players in that race are pretty much set now, Warner and Kant argue. And poolside, which has raised $626 million and works with a cluster of 10,000-plus GPUs as of the fall, would seemingly face long odds to leapfrog to the front.
But Warner and Kant say they have the world’s largest code execution environment with which to train their agents. “It only takes one breakthrough to turn the tables on everyone,” says Kant.
Highlights of Upstarts’ interviews with the poolside founders can be found below, including:
Why VCs self-interest is exaggerating an AI opportunity
Why vibe-coding tools might have a brighter future than code editors
How AI agents are already changing jobs inside poolside
Meta’s “worst sin imaginable” behind the Scale AI deal