TechCrunch Disrupt 2026: Ricursive Intelligence’s Anna Goldie and Azalia Mirhoseini on when AI starts designing its own hardware
TechCrunch TechCrunch Events
Ricursive wants AI to help design chips, cutting the process from years to weeks. Its founders think faster chip design could speed up AI itself.
Based on reporting by TechCrunch, TechCrunch Events — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
AI keeps getting more demanding, and the chips underneath it are not keeping pace. Designing those chips is slow, intricate work that can take two to three years. Ricursive Intelligence is betting that AI can help break that logjam by doing part of the design work itself, then learning from each chip to improve the next one.
The company’s founders, Anna Goldie and Azalia Mirhoseini, are not starting from scratch. Before Ricursive, they co-led AlphaChip at Google, an AI system that generated chip layouts in hours instead of the much longer process used by human designers. That work helped shape multiple generations of Google’s Tensor Processing Units. Now they want to push the idea further, turning chip design into a feedback loop rather than a one-off engineering slog.
Ricursive says its system is built to learn across different chips, so each design informs the next. The goal is to automate more of the process, including component placement and design verification, while opening the door to new chip architectures. If that works, the payoff is bigger than speed. Better chip design could mean better hardware for AI, which in turn could help create even more capable systems.
The company has attracted money fast. Ricursive launched in late 2025 and raised $335 million within four months at a $4 billion valuation. That round included a $300 million Series A, and Nvidia is among the investors. Goldie is Ricursive’s CEO; Mirhoseini is CTO. Goldie also has a Ph.D. from Stanford and was named one of MIT Technology Review’s 35 Innovators Under 35, while Mirhoseini is a Stanford assistant professor and founder of its Scaling Intelligence Lab.
Goldie and Mirhoseini will talk about all of this at TechCrunch Disrupt 2026 on the Disrupt Stage in a session called “When AI Starts Designing Its Own Hardware.” The broader point is simple: AI may be moving fast, but its hardware cycle is still stuck in old-school time. Ricursive is trying to shorten that gap, and that could matter as much as any model upgrade.
My take — AI-written commentary, not fact-checked reporting
Ricursive is chasing the kind of problem that actually matters: the boring, expensive bottleneck nobody can tweet into existence. The hype machine loves model tricks; the real leverage is in hardware that arrives before the next wave of models is already outdated. If AI is going to eat the world, someone has to teach it to build the kitchen.
Read more about this at: TechCrunch