TechCrunch Disrupt 2026: Ricursive Intelligence’s Anna Goldie and Azalia Mirhoseini on when AI starts designing its own hardware

AI needs increasingly powerful chips. But designing those chips is itself a slow, complex process. What happens when AI starts doing some of that work, too?

That’s the idea behind Ricursive Intelligence. The startup is developing AI that can design chips, learn from the process, and improve the next design — potentially accelerating development of the hardware that powers increasingly capable AI.

Today, designing a chip can take two to three years. Recursive wants to reduce that cycle to a matter of weeks.

At TechCrunch Disrupt 2026, Ricursive Intelligence co-founders Anna Goldie and Azalia Mirhoseini will take the Disrupt Stage for “When AI Starts Designing Its Own Hardware.” They’ll discuss closing the loop between AI and chip development, building self-improving systems, and tackling one of the biggest bottlenecks in the race toward more powerful intelligence.

Want to hear from the researchers working to change how AI hardware gets built? Today is the last day to save up to $200 on your Disrupt pass. Secure your ticket by 11:59 p.m. PT — and get a second pass for 50% off select ticket types.

What happens when AI designs the chips that power it?

Goldie and Mirhoseini have already shown that AI can play a significant role in chip design. Before founding Ricursive Intelligence, the pair co-led AlphaChip at Google, an AI system that could generate chip layouts in hours rather than the much longer process required of human designers. Their work helped design multiple generations of Google’s Tensor Processing Units.

Ricursive Intelligence is taking that concept further, building AI tools to automate and accelerate chip design. Its system is intended to learn across different chips, so the experience of designing one can improve how it approaches the next.

That creates a potentially powerful loop: AI helps design better hardware, better hardware supports more powerful AI, and those systems can then help develop what comes next.