Applied Compute

Applied Compute runs the Applied Compute Agent Cloud (AC2), a platform that lets companies train, serve, and continuously improve their own custom AI models instead of relying entirely on third-party model APIs. It covers the full loop: training specialized models on a client's proprietary data with post-training optimization, serving them in production-grade inference infrastructure, and improving them over time using real production feedback. The pitch, echoed by Microsoft CEO Satya Nadella in a fireside chat featured on the company's own site, is that "there should be as many models in the world as there are firms" — every serious company should be able to own a model tuned to its exact use case, not just call a generic API. Its named customers include Cognition, Microsoft, NVIDIA, Harvey, DoorDash, NTT Data, Mercor, and Latch Bio; Harvey specifically credits it with enabling large-scale reinforcement learning inside a custom legal-agent system. Applied Compute is SOC 2 and ISO 42001 certified and offers optional VPC deployment for enterprise data control. It was founded by Yash Patil (CEO), Rhythm Garg (CTO), and Linden Li (Chief Architect), and has raised a $20M Series A (June 2025) followed by two Series B tranches — $60M in October 2025 and $80M in April 2026, the latter led by Kleiner Perkins.