The co-founders of the startup Ricursive Intelligence seemed destined to be partners. Anna Goldie, the CEO, and Azalia Mirhoseini, the CTO, are so well-known in the AI community that they were among the engineers who received unusual emails from Mark Zuckerberg making extravagant offers. They did not accept those offers. The pair worked together at Google Brain and were early employees at Anthropic.
They earned acclaim at Google by creating the Alpha Chip, an AI tool that could generate solid chip layouts in hours—a process that normally takes human designers a year or more. This tool helped design three generations of Google’s Tensor Processing Units.
That pedigree explains why, just four months after launching Ricursive, they announced a $300 million Series A funding round at a $4 billion valuation led by Lightspeed. This came only a couple of months after raising a $35 million seed round led by Sequoia.
Ricursive is building AI tools that design chips, not the chips themselves. This makes them fundamentally different from nearly every other AI chip startup; they are not a would-be competitor to Nvidia. In fact, Nvidia is an investor. The GPU giant, along with AMD, Intel, and every other chip maker, are the startup’s target customers.
The founders stated their goal is to enable any kind of chip to be built in an automated and accelerated way using AI. Their paths first crossed at Stanford, where Goldie earned her PhD while Mirhoseini taught computer science classes. Since then, their careers have moved in lockstep. They started at Google Brain on the same day, left on the same day, joined Anthropic on the same day, left on the same day, rejoined Google on the same day, left again on the same day, and finally started their company together on the same day.
During their time at Google, the colleagues were so close they even worked out together, both enjoying circuit training. The pun was not lost on Jeff Dean, the famed Google engineer who collaborated with them. He nicknamed their Alpha Chip project “chip circuit training,” a play on their shared workout routine. Internally, the pair also got the nickname A&A.
The Alpha Chip earned them industry notice, but it also attracted controversy. In 2022, a colleague at Google was fired after spending years trying to discredit A&A and their chip work, even though that work was used to help produce some of Google’s most important AI chips.
Their Alpha Chip project at Google Brain proved the concept that would become Ricursive: using AI to dramatically accelerate chip design. The issue is that computer chips have millions to billions of logic gate components integrated on their silicon wafer. Human designers can spend a year or more placing those components to ensure performance and good power utilization. Digitally determining the placement of such infinitesimally small components with precision is extremely difficult.
The Alpha Chip could generate a very high-quality layout in about six hours. The founders explain that the approach learns from experience. It uses a reward signal that rates how good a design is, and the AI agent uses that rating to update its deep neural network to improve. After completing thousands of designs, the agent became very good and also got faster as it learned.
Ricursive’s platform aims to take this concept further. The AI chip designer they are building will learn across different chips, so each design helps it become better for the next one. Their platform also makes use of large language models and will handle everything from component placement through design verification. Any company that makes electronics and needs chips is their target customer.
If their platform proves itself, Ricursive could play a role in the moonshot goal of achieving artificial general intelligence. Their ultimate vision is designing AI chips, meaning the AI will essentially design its own computer brains. They believe building more powerful chips is the best way to advance the frontier of AI. They also argue that the lengthy chip-design process constrains how quickly AI can advance, and their work could enable a fast co-evolution of AI models and the chips that power them.
If the thought of AI designing its own brains brings visions of science fiction to mind, the founders point to a more immediate and likely benefit: hardware efficiency. When AI labs can design far more efficient chips, their growth will not have to consume so many of the world’s resources. They claim the potential for almost a tenfold improvement in performance per total cost of ownership.
While the young startup will not name its early customers, the founders say they have heard from every major chip-making company imaginable. Unsurprisingly, they have their pick of first development partners.