Ex-DeepMind researcher raises $55M seed round at $300M valuation for visual AI

Andrew Dai spent a decade at Google DeepMind working on research that would later inform systems like ChatGPT. But when he left to start his own company, he didn’t chase the obvious prize. Instead of large language models, Dai bet on visual AI — an area he describes as the next major frontier where progress has been painfully uneven.

That bet is paying off faster than most seed-stage gambles. His startup, Elorian, just closed a $55 million seed round at a $300 million valuation, a ratio that outpaced even Thinking Machines’ historic U.S. raise. Dai pulled it off just months after walking away from Big Tech, a timeline that speaks to both his pedigree and the ravenous investor appetite for frontier AI teams.

On a recent episode of Build Mode, Dai told host Isabelle Johannessen why visual understanding remains AI’s hard problem. “You have models that are doing really great at math, really great at new physics ideas, and of course coding is very popular now,” he said. “But one area where progress has been extremely uneven is visual understanding and visual reasoning.” Elorian’s pitch is refreshingly direct: build models that push toward visual AGI.

The fundraising process itself offers a clinic in founder discipline. Dai fielded offers at even loftier valuations but chose Nvidia and Menlo Ventures as strategic partners instead. His reasoning? Investors who actually understand what it takes to build frontier AI are worth more than a bigger number on a term sheet. The conversation also digs into practical mechanics — how to pitch deeply technical products without drowning non-technical VCs in jargon, why speed has become the single biggest competitive moat in AI, and the art of luring world-class researchers away from the comfort of Big Tech salaries. New episodes of Build Mode drop every Thursday across podcast platforms and YouTube.