Flapping Airplanes raises a $180M seed at $1.5B

The lab was co-founded by brothers Ben and Asher Spector with Aidan Smith, and it is building models that need far less data and compute than current approaches. Menlo Ventures, Andrej Karpathy, Jeff Dean and Chris Re also took part in the round. No commercial product has launched, and the pitch to investors rests on a research-first approach. The stated aim is to improve training data efficiency by 1000 times, using the human brain as inspiration rather than scaling existing architectures. The company is part of a group of so-called neolabs, startup research labs that investors are backing ahead of any revenue. Sequoia partner David Cahn frames it as a shift toward long-horizon research, arguing that a few core scientific advances could unlock major progress toward advanced intelligence. The Wall Street Journal described these labs as drawing intense interest from investors. For founders, the round shows that capital is available for teams that challenge the scale-first approach, but it is going to credentialed researchers with strong investor networks and a narrow technical thesis. Labs that depend on ever larger data and compute budgets now face a well-funded competitor arguing the opposite. Application-layer startups should watch the claims on data efficiency, since any real gain would lower training costs and shift where the advantage sits.