Flapping Airplanes
Category: Foundation Models / LLMs
A foundational AI research lab (Neo Lab) developing next-generation model architectures designed to achieve intelligence through data efficiency rather than brute-force scaling. Flapping Airplanes was founded in 2025. The company is led by Ben Spector. Based in San Francisco, USA. Team size: 10-50. Total funding raised: $180.0M. Latest round: Seed. Key investors include GV (Google Ventures), Sequoia Capital, Index Ventures, Menlo Ventures, XTX Ventures, Fundomo, Conviction Investment Partners, Nova Global, Victor Lazarte, Lux Capital, Argo, Jump Trading.
- Founded
- 2025
- Headquarters
- San Francisco, USA
- Team size
- 10-50
- Total funding
- $180.0M
Value proposition
Challenges the current 'scaling laws' by discovering the fundamental principles of learning, enabling models to reach high performance with significantly less data and compute.
Products and solutions
Next-generation neural architectures (Post-Transformer research), Data-efficient training frameworks, High-performance ML systems optimization, Sample-efficient learning algorithms
Unique value
The company operates on the 'fixed-wing' flight metaphor: just as aviation moved from imitating birds (flapping) to understanding aerodynamics, they aim to move from imitating human-scale data consumption to understanding the mathematical principles of intelligence.
Target customer
Foundational model developers, enterprise AI researchers, and organizations requiring high-performance AI without the prohibitive costs of massive-scale data ingestion.
Industries served
AI Infrastructure, Enterprise Software, Research & Development, High-Performance Computing
Technology advantage
Led by the co-creator of Flash Attention, the team possesses world-class expertise in ML systems and hardware optimization, allowing them to build architectures that are natively optimized for modern compute environments.
How they differentiate
Focuses on 'fixed-wing' intelligence—discovering the fundamental mathematical principles of learning to achieve high performance with minimal data, rather than the 'flapping' approach of brute-force scaling and ingesting the entire internet.
Main competitors
Sakana AI, Physical Intelligence (π), World Labs, Mistral AI
Key partnerships
Stanford Hazy Research Lab (Academic lineage)
Major milestones
Founded in 2024 by Flash Attention co-creator Ben Spector, Secured $180M Series A funding in February 2025, Assembled a founding team of researchers from Stanford, MIT, and Neuralink
Growth metrics
Pre-product research stage; high-capital intensity for compute acquisition and elite talent recruitment.
Market positioning
Foundational AI research lab (Neo Lab) challenging the Transformer-based scaling dogma.
Geographic focus
North America (San Francisco), Global Research Community
Patents and IP
No public patent filings discovered as of early 2025; currently operating with proprietary research and trade secrets.
About Ben Spector
Ben Spector is a Stanford Computer Science PhD student (currently on leave) and a 2023 Hertz Fellow. He was a prominent researcher in the Hazy Research lab under Chris Ré, where his work focused on the intersection of machine learning and systems. He is a key contributor to the development of Flash Attention, a widely adopted optimization for Transformers, and has extensive expertise in high-performance ML training and AI hardware optimization.
Latest news about Flapping Airplanes
More Foundation Models / LLMs companies
Official website: https://flappingairplanes.com/