Ricursive Intelligence's founders on AI that designs its own chips

The semiconductor industry faces a fundamental constraint: the very chips that power advanced AI systems take years to design. Ricursive Intelligence, a startup founded by former Google researchers, is tackling this bottleneck by using AI to accelerate the chip design process itself.

The company's premise is straightforward but ambitious. Rather than relying solely on human engineers, Ricursive is developing AI systems that can generate chip layouts, learn from each design cycle, and apply those lessons to the next generation. The potential payoff is significant: compressing what currently takes two to three years into a matter of weeks.

From research to startup

Co-founders Anna Goldie and Azalia Mirhoseini bring substantial credentials to the venture. Before launching Ricursive in late 2025, both held senior positions at Google DeepMind and spent time as early employees at Anthropic. They also co-led AlphaChip at Google, an AI system that demonstrated the feasibility of using machine learning to generate chip layouts in hours rather than the traditional manual approach.

Goldie serves as founder and CEO and holds a Ph.D. in computer science from Stanford; she was recognized as one of MIT Technology Review's 35 Innovators Under 35. Mirhoseini, the company's founder and CTO, is an assistant professor of computer science at Stanford and established the university's Scaling Intelligence Lab.

The startup attracted investor attention quickly. Within four months of its founding, Ricursive raised $335 million at a $4 billion valuation, with a $300 million Series A round. Nvidia is among the investors backing the company.

Closing the loop between AI and hardware

Ricursive's approach aims to automate multiple stages of chip design, from component placement through design verification. The system is built to retain knowledge across different chip projects, allowing insights from one design to inform subsequent ones.

This creates what Goldie and Mirhoseini see as a virtuous cycle: AI helps design superior hardware, superior hardware enables more capable AI systems, and those systems can then contribute to the next iteration of chip development. By accelerating the pace of chip design, the founders believe they can unlock new chip architectures and ultimately deliver AI systems that are both more powerful and more efficient.

The hardware-software timing problem

AI model capabilities have expanded at a rapid pace, but the hardware supporting those models operates on a slower development timeline. Ricursive is betting that AI-assisted design can help synchronize these two timelines.

Goldie and Mirhoseini are scheduled to discuss their work at TechCrunch Disrupt 2026, taking the stage to explore when AI begins designing the hardware that powers it. The session will cover the technical approach, the industry bottlenecks they are targeting, and how self-improving systems might reshape chip development. The event runs October 13–15 at Moscone West in San Francisco.