A 25-year-old Stanford PhD student who never pitched a product just closed $180 million for an AI lab that doesn't plan to ship anything soon. Ben Spector's Flapping Airplanes raised the round from GV, Sequoia Capital, Index Ventures, and Menlo Ventures at a $1.5 billion valuation, according to a Wall Street Journal report this week.
The premise is simple, if audacious: current AI models are wildly inefficient. Humans pick up language and reasoning from maybe a few billion tokens over a lifetime. GPT-4 trained on trillions. Spector thinks that gap, somewhere between 100,000x and a million times, represents an engineering problem worth solving.
The team recruiting kids over PhDs
Spector isn't chasing big-name researchers. The 11-person team includes his brother Asher, 26, who just finished a statistics PhD at Stanford, and co-founder Aidan Smith, a 21-year-old Thiel Fellow who spent three years at Neuralink while commuting to Georgia Tech. There's also an 18-year-old high schooler on staff.
The hiring thesis comes straight from Ben's background running Prod, a nonprofit accelerator he founded at MIT. Prod alumni include the founders of Cursor (now valued at roughly $30 billion), Mercor, and Etched. According to Index Ventures, the accelerator's portfolio companies have raised over $25 million collectively.
Sequoia partner David Cahn, who led part of the round, called it a bet on raw talent over credentials. In a blog post announcing the investment, he cited Einstein's 26th year as evidence that fundamental breakthroughs come from the young. Whether that logic scales to AI research, where compute access and institutional knowledge matter enormously, remains an open question.
Why the name matters
Flapping Airplanes references a debate in early aviation: should we copy birds or build something new? The Wright brothers won by ignoring biology. But Spector's argument runs the other direction. He thinks current AI architectures are the fixed-wing approach, and that biological learning contains something worth studying.
The company isn't committed to any single technical path. That's either intellectual honesty or a red flag, depending on your tolerance for open-ended research bets. Cahn frames it as exploring "new architectures, loss functions, and even gradient descent" itself.
The neolab moment
Flapping Airplanes joins a growing category some call neolabs, research-first AI startups raising venture money without product roadmaps. Safe Superintelligence, founded by former OpenAI chief scientist Ilya Sutskever, has raised around $3 billion at a $32 billion valuation. Reflection AI pulled in $2 billion. Humans& raised $480 million in January.
The pitch is straightforward: OpenAI started as a research lab before becoming a company worth over $150 billion. Investors want exposure to that trajectory without waiting for products to materialize. Critics suggest most neolabs will produce incremental research at best. The talent retention problem is real. Thinking Machines Lab, co-founded by former OpenAI executive Mira Murati, lost three co-founders to Big Tech in recent months despite targeting a $50 billion valuation.
Flapping Airplanes has lined up Andrej Karpathy as an advisor and Jeff Dean as an angel investor, names that signal credibility within the research community. Whether that translates to breakthrough results is a different question entirely.
What happens next
The company is hiring at [email protected]. Ben Spector took leave from Stanford's PhD program in September. The lab has a Hugging Face organization page, though it's sparse.
There's no timeline for publications, no benchmark targets announced, and no product plans. For a company predicated on making AI development more efficient, the pitch is remarkably patient: give smart young people freedom, fund them generously, and wait.
