Year-old AI lab Flapping Airplanes raises $1B at a $6B valuation

You can now train AI models with significantly less data, potentially lowering costs for businesses.

Flapping Airplanes, an AI laboratory founded nine months ago, has secured $1 billion in Series A funding at a $6 billion valuation. The round was co-led by Index and Lux, with participation from previous investors like GV, Sequoia, and Kleiner Perkins, as well as new investors such as Valor, Atreides, A*, and Nvidia. The company previously raised a $180 million seed round at a $1.5 billion valuation. Founded by Aidan Smith, Asher Spector, and Ben Spector, Flapping Airplanes aims to train large AI models using significantly less data. They have developed a "megakernel framework" that can perform training, inference, and other tasks simultaneously within a single kernel, reportedly keeping GPUs from being idle and reducing the cost and data requirements for AI systems. The founders claim this approach can lead to more data-efficient models that are easier to integrate into the economy.

Flapping Airplanes' rapid fundraising success and ambitious valuation highlight a strong investor appetite for AI companies focused on efficiency and cost reduction. The company's "megakernel framework" addresses a key industry challenge: the immense computational resources and data required to train and run large AI models. By enabling simultaneous training and inference within a single kernel, Flapping Airplanes suggests a path towards more accessible and economical AI development.

This development could signal a shift in AI development priorities, moving beyond simply scaling up models to optimizing their resource utilization. If successful, Flapping Airplanes' technology might disrupt the current paradigm of data-intensive AI training, potentially impacting hardware manufacturers and cloud providers by changing the demand for raw compute power and storage. Investors will be watching to see if the company can deliver on its promise of drastically reduced training costs and data needs.