A Young AI Lab Just Raised $1 Billion to Make AI Learn From Less Data

Flapping Airplanes announced on Wednesday, October 21, 2026, that it has raised $1 billion in a funding round co-led by Index and Lux, according to TechCrunch. Bloomberg reported that the round values the AI research startup at $6 billion, in a report published the same day. Quartz noted that the valuation is four times what the company was worth when it launched in January 2026.

Sequoia, GV, and Kleiner Perkins took part in the round. Valor, Atreides, Kleiner Perkins, A*, and Nvidia are also listed among the company's investors.

The company launched in January 2026 with a $180 million seed round from GV, Sequoia, and Index. That round valued it at $1.5 billion. TechCrunch's January 29 coverage called the seed the debut of a new AI lab. Aidan Smith and brothers Asher Spector and Ben Spector founded the company.

The company calls itself a "frontier data efficiency lab." It is still in stealth, which means it has not shared much about its work publicly. Its website says humans are 100,000x to 1,000,000x more efficient learners than existing AI models. Think of a child who learns a new word after hearing it a few times. Today's AI models usually need far more examples than that. The company's goal is to train large AI models on less data.

It also claims its software, called a megakernel framework, lets two kinds of AI work share the same chips at once instead of waiting in line. The company says it is not trying to sell anything yet and is focused on research. Its website lists team members who won medals in international math, computing, and physics competitions. It also names Chris Re, Andrej Karpathy, and Jeff Dean among its angels and advisors.

Training AI takes huge amounts of data and computing time. If the efficiency claim worked at scale, each new skill would need less data, which would lower training costs more directly than buying more computers would. Running and training AI on the same chips is hard, because the two jobs stress the chips in opposite ways. The company describes both claims as its research direction, not as proven results.

In my view, the gap between those claims and the money raised is the thing to watch. A figure like 100,000x to 1,000,000x is only a hypothesis until independent researchers test it with methods others can repeat. A $6 billion valuation for a lab that is not selling anything places a high price on results that do not yet exist. Nvidia, which makes the chips AI training runs on, has a clear stake in whether efficiency gains lower or raise demand for those chips. That question is still open.

The broader context here is that the biggest funding rounds increasingly go to labs built around one research idea instead of a product plan. If the idea works, the benefit could reach anyone building large AI models, not only this company.

The main point is simple. A young lab founded by three researchers now has $1 billion to test whether its claims hold up under measurement.