Ricursive Intelligence

Ricursive Intelligence is a frontier AI lab working on self-improving AI systems, beginning with chip design. The company applies artificial intelligence to chip architecture, verification and layout, aiming to close the loop between AI and the hardware it runs on. Its technical work spans reinforcement learning, AI hardware co-design and chip design automation, alongside research into artificial superintelligence.

The lab's research output includes four breakthrough papers. AlphaChip, on AI-driven chip design, was published in Nature in 2021. RL-CCD received a Best Paper award at DAC 2023, Insta at DAC 2025, and C3PO at ASP-DAC 2026. Alongside these, the company is developing a recursive chip design system, a shift from sequential chip design toward a self-improving loop in which AI designs better chips that in turn enable better AI models.

The company has raised $335M and is a small team recruited from institutions including Google DeepMind, Anthropic, NVIDIA, Cadence, Apple, xAI, Stanford, MIT and Harvard, with hands-on experience building frontier models and hardware such as Gemini, Claude, Grok and TPUs. The stated focus is on the most important bottlenecks in AI and chip design.