News
eBook
Designing an AI-Native Contact Center
Key Takeaways
Ricursive Intelligence raised $300 million at a $4 billion valuation.
The startup aims to close the feedback loop between AI models and the silicon they run on.
As hyperscalers race for custom silicon, faster chip innovation is becoming a competitive advantage.
A frontier AI lab planning to close the loop between artificial intelligence and chip design just raised $300 million at a $4 billion valuation.
Ricursive Intelligence announced on Jan. 26, 2026, that it closed a Series A round less than two months after launching with a $35 million seed at a $750 million valuation.
Lightspeed Venture Partners led the round, with participation from DST Global, NVentures (Nvidia's venture capital arm), Felicis Ventures, 49 Palms, Radical AI and Sequoia Capital. Total funding now stands at $335 million.
According to company officials, Ricursive is building a platform "that closes the recursive feedback loop between AI and the chips that power it." The company claims its approach allows AI to "continuously improve the silicon it depends on, creating a self-reinforcing cycle of advancement."
Table of Contents
From AlphaChip to a New AI Lab
"Chips are the fuel for progress in AI, and the multi-year chip design process is holding back the field."
- Dr. Anna Goldie
Co-Founder & CEO, Ricursive Intelligence
Ricursive Intelligence debuted on Dec. 2, 2025, announcing a $35 million seed round led by Sequoia Capital. Co-founders Dr. Anna Goldie (CEO) and Dr. Azalia Mirhoseini (CTO), former Google researchers who created AlphaChip, positioned the lab as a challenger to traditional semiconductor design workflows.
"Chips are the fuel for progress in AI, and the multi-year chip design process is holding back the field," Dr. Goldie wrote on LinkedIn. "At Ricursive Intelligence, we’re using AI to revolutionize chip design, developing superior hardware in a fraction of the time. By closing the loop between AI and hardware, we can accelerate progress toward artificial superintelligence and enable a Cambrian explosion of custom silicon."
The first phase of the company is to accelerate chip design, noted Dr. Goldie. "But if we have the ability to design chips very quickly, why not just use that ourselves? Why not build our own chips? Why not train our own models? Why not co-evolve them?"
AI Compute Hits a Breaking Point
Major hyperscalers have moved decisively to develop proprietary chips, all aiming to control their AI compute destiny:
- Meta acquired Rivos to build custom AI accelerators
- Amazon developed AWS Trainium and Inferentia
- Microsoft created Azure Maia and Cobalt
- Google built Axion
- Apple announced its M5 chip
Performance Gains Break the Old Rules
AI supercomputer performance is doubling every nine months, a pace that rivals Moore's Law. AMD's Instinct MI350 series, built on TSMC's 3 nm process, introduced new low-precision formats like FP4 and FP6.
AMD silicon powers the world's fastest supercomputers, with Frontier and El Capitan holding the #1 and #2 slots.
According to industry experts, AI hardware innovation now depends more on architectural breakthroughs than transistor density. Extropic's probabilistic computing approach, for example, claims up to 10,000x energy savings compared to traditional GPU-based systems.
About Ricursive Intelligence
Ricursive Intelligence is an AI research and engineering company founded in 2025 in Palo Alto, focused on automating and accelerating chip design for hyperscalers, cloud AI providers and semiconductor firms. The company applies machine learning to automate chip design tasks such as floorplanning and layout.