Big Tech AI Researchers Are Leaving to Launch Billion-Dollar Labs, and VCs Are Writing the Checks

The biggest AI labs in the world are bleeding their best researchers to startups, and investors are funding the departures at record scale. VCs have channeled $18.8 billion into AI startups founded since the start of 2025, according to Dealroom data cited by CNBC, on pace to exceed the $27.9 billion raised last year by companies launched since 2024.

The numbers behind individual departures are staggering. David Silver, the DeepMind researcher behind AlphaGo, closed a $1.1 billion seed round for Ineffable Intelligence. Tim Rocktäschel, also formerly of DeepMind, is reportedly raising up to $1 billion for Recursive Superintelligence, according to the Financial Times. AMI Labs, founded by former Meta AI chief Yann LeCun, announced a $1 billion raise in March. Former staff at OpenAI, Anthropic, and xAI have raised hundreds of millions more for ventures including Ricursive Intelligence ($335 million), Periodic Labs ($300 million), and Humans&.

Why Researchers Are Leaving

The pattern is consistent: researchers cite narrowing focus and commercial pressure at their former employers as the catalyst.

“When you’re in a race, you narrow focus,” Elise Stern, managing director at French VC Eurazeo, told CNBC. “That creates a vacuum. Entire areas of research, like new architectures, agents, interpretability and vertical models, are being deprioritised, not because they don’t matter, but because they don’t win the immediate race.”

Alexander Joël-Carbonell, partner at HV Capital, made a similar point: “Inside the large foundational labs, the pressure to deliver benchmark performance and maintain rapid release cycles leaves limited room for genuinely exploratory research, particularly outside the dominant LLM paradigm.”

Ricursive Intelligence co-founder Anna Goldie, who previously worked at both Anthropic and DeepMind, told CNBC that independence offered a competitive advantage with customers. “For chipmakers to trust us with their most valuable IP, we have to be Switzerland, and that wouldn’t be possible if we were at Google,” she said.

The Funding Flywheel

The departures feed on themselves. New labs recruit from their founders’ former employers, creating a talent drain that compounds over time. Ricursive Intelligence reassembled the core AlphaChip team from DeepMind. Sequoia Capital, Google, and NVIDIA are all backing multiple new entrants.

The investor thesis, according to CNBC, is that foundational AI research will concentrate value in smaller labs rather than Big Tech R&D divisions. Eurazeo’s Stern framed it directly: former lab researchers “know what works at scale, and they know exactly what is being left on the table internally.”

The Competitive Cost for Big Tech

The exodus raises a structural question for Google, Meta, and OpenAI. These companies are spending hundreds of billions on AI infrastructure, but the researchers who built their foundational capabilities are leaving to compete with them. The new labs focus on reinforcement learning, autonomous reasoning, chip design, and agent systems: precisely the areas Big Tech needs for the next generation of products.

For agent builders, the implication is a broadening supply side. More independent labs pursuing foundational research means more diverse models, more specialized capabilities, and more competitive pricing over time. The question is whether Big Tech can retain enough talent to maintain its lead, or whether the startup wave forces a shift in how foundational AI research gets funded and commercialized.