San Francisco-based Flapping Airplanes has raised $1 billion in Series A funding co-led by Index Ventures and Lux Capital, at a reported $6 billion valuation, despite having no commercially released AI model.
The financing comes approximately nine months after the research startup announced a $180 million seed round backed by investors including GV, Sequoia Capital and Index Ventures. That round valued the company at approximately $1.5 billion, meaning its reported valuation has quadrupled within a year.
Other investors associated with the company include Kleiner Perkins, Valor, Atreides Management, A* and Nvidia.
Flapping Airplanes was founded by brothers Ben and Asher Spector and Aidan Smith. Ben previously studied computer science at Stanford and founded Prod, a student-run startup accelerator. Asher pursued doctoral research in statistics at Stanford, while Smith previously worked at Neuralink and was a Thiel Fellow.
The company is pursuing a different approach to the increasingly expensive development of frontier AI systems.
Most leading laboratories have focused on improving model performance through larger training datasets, more computational resources and increasingly sophisticated training methods.
Flapping Airplanes believes substantial efficiency gains remain possible by rethinking how AI models interact with computing hardware and learn from data.
Its research spans model architecture, pre-training, post-training, inference and GPU programming.
One of its central technologies is a framework for megakernels, GPU programs that combine operations normally executed through multiple separate computational steps.
Traditional AI workloads often involve launching many individual GPU kernels, each performing a particular operation. Moving between those operations can introduce synchronization delays, data movement and hardware inefficiencies.
Flapping Airplanes is investigating whether more of the AI workload can be executed within a single coordinated GPU program.
According to the company, its framework can combine model training and inference, potentially allowing memory-intensive and compute-intensive operations to use different resources on the same GPU simultaneously.
The founders argue that existing AI systems do not fully exploit the capabilities of today's computing hardware.
If successful, such techniques could lower the computational cost of training and deploying models without requiring equivalent improvements in semiconductor manufacturing.
But hardware optimization is only part of the company's ambition.
Flapping Airplanes also wants to make AI models substantially more data-efficient, reducing the amount of information needed to develop highly capable systems.
The approach could be relevant to industries where training data is scarce or expensive to obtain, including robotics, scientific research and specialized industrial applications.
The company currently has a relatively small research team of approximately two dozen employees, according to Index Ventures. It operates in person in San Francisco and is building expertise across the entire AI stack rather than concentrating on a single product.
The new capital will support research, recruiting and the computing infrastructure required to test its ideas at larger scales.
The financing reflects investor confidence in the possibility that the next major advance in AI may come from improving efficiency rather than simply increasing spending.
However, Flapping Airplanes remains an early-stage research organization.
Its reported $6 billion valuation has not been independently confirmed through a company disclosure, and it has not publicly released a commercial product, customer revenue figures or comprehensive independent benchmarks demonstrating its claimed efficiency advantages.
The company also faces competition from established AI laboratories and specialized inference infrastructure businesses pursuing similar improvements.
The key question is whether its research can deliver meaningful reductions in training and inference costs at production scale.
The $1 billion financing gives Flapping Airplanes resources comparable to much more mature companies. Its challenge is turning a promising technical thesis into measurable breakthroughs and, eventually, a commercially viable AI platform.