Ricursive Intelligence

Ricursive Intelligence is a frontier AI lab headquartered in Palo Alto, California, that develops AI systems for semiconductor design. The company was founded by Dr. Anna Goldie (CEO) and Dr. Azalia Mirhoseini (CTO), the co-creators of AlphaChip, the reinforcement-learning chip-floorplanning system published in Nature and used across four generations of Google TPUs and by external semiconductor companies. Ricursive's stated goal is to close a recursive feedback loop between AI models and the chips that run them, treating the slow, capital-intensive semiconductor design process as the primary bottleneck to AI progress.

The company builds AI tooling that designs chips rather than the chips themselves; per TechCrunch, this positions it as a supplier to chip makers such as Nvidia, AMD and Intel rather than a competitor to them. Its platform is intended to span the design stack from component placement through design verification, incorporating large language models alongside reinforcement learning, and to learn across successive chip projects so that each design improves the designer. Published commentary describes a phased roadmap: first compressing the multi-year design cycle for the hardest stages of chip design, then end-to-end automation from workload specification to GDS2 output for foundries, and eventually vertical integration in which Ricursive designs its own chips and trains its own models on them.

The team draws on researchers and engineers from Google DeepMind, Anthropic, NVIDIA, Cadence, Apple, xAI, Stanford, MIT and Harvard, and the company cites authorship of AlphaChip (Nature 2021), RL-CCD (DAC Best Paper 2023), Insta (DAC Best Paper 2025) and C3PO (ASP-DAC Best Paper 2026), as well as hands-on work on Gemini, Claude, Grok and TPUs.

Founding story

Goldie and Mirhoseini first crossed paths at Stanford, where Goldie completed a PhD in NLP while Mirhoseini taught computer science. Their careers ran in parallel: they started at Google Brain on the same day, joined and left Anthropic on the same day, rejoined and left Google on the same day, and then founded Ricursive together. At Google Brain they co-created AlphaChip, an approach that reportedly began when both were independently drafting a moonshot proposal for AI-driven chip design and Jeff Dean emailed them the same idea; Dean nicknamed the effort 'chip circuit training' after the pair's shared workout routine. AlphaChip's demonstrated ability to produce high-quality layouts in hours rather than the year or more typically needed by human designers became the proof of concept for the company.

Business model

Ricursive develops an AI platform that automates semiconductor design for chip makers and other companies that need custom or traditional silicon; specific pricing or contracting terms are not disclosed in the available sources. The company's longer-term stated plan includes vertical integration into designing its own chips and training its own models.

Traction

Since launch the company has recruited researchers and engineers from Google DeepMind, Anthropic, Apple and Cadence, including several who worked directly on AlphaChip. It raised a $300 million Series A at a $4 billion post-money valuation within two months of public launch, and the founders claim potential performance-per-total-cost-of-ownership gains of roughly 10x from co-designing architectures with models. No customers or revenue have been disclosed.

Latest developments

On 26 January 2026 the company announced a $300 million Series A led by Lightspeed Venture Partners at a $4 billion post-money valuation, with participation from DST Global, NVentures, Felicis Ventures, 49 Palms Ventures, Radical AI and Sequoia Capital; proceeds are earmarked for team growth and expanded compute infrastructure. A February 2026 TechCrunch feature put cumulative funding at $335 million raised in four months. The company's site cites $335M in backing from Sequoia, Lightspeed, DST and NVentures and lists numerous open Palo Alto roles.