Ricursive Intelligence has quickly become one of the more closely watched startups in AI hardware, but its bet is not to build another AI chip. Its plan is to build AI that helps design chips faster, from component placement through design verification.
That distinction matters. Instead of trying to compete with Nvidia, AMD, Intel, and other chipmakers, Ricursive wants to sell tools to them and to any company that makes electronics and needs chips.
A fast rise built on a long track record
Ricursive was founded by Anna Goldie, CEO, and Azalia Mirhoseini, CTO. Both are well known in the AI community, and their resumes explain why investors moved quickly.
The two worked together at Google Brain and were early employees at Anthropic. At Google, they created Alpha Chip, an AI tool that could produce solid chip layouts in hours. The source article says the same kind of work normally takes human designers a year or more.
Alpha Chip helped design three generations of Google’s Tensor Processing Units. That history gave Ricursive a clear proof point before the startup had much time in the market.
Four months after launching, Ricursive announced a $300 million Series A round at a $4 billion valuation led by Lightspeed. That followed a $35 million seed round led by Sequoia just a couple of months earlier. Together, the rounds brought in $335 million.
What Ricursive is actually building
Ricursive is not making its own chips. It is building AI tools that design chips. That makes the company different from many AI chip startups, because it is not positioning itself as a wannabe Nvidia competitor.
Nvidia is actually an investor. The source article identifies Nvidia, AMD, Intel, and every other chip maker as part of Ricursive’s target customer universe.
Mirhoseini described the ambition this way: “We want to enable any chip, like a custom chip or a more traditional chip, any kind of chip, to be built in an automated and very accelerated way. We’re using AI to do that,”
The platform is meant to handle more than one step of chip work. According to the source, Ricursive’s system will use LLMs and cover tasks from component placement through design verification.
That gives the company a broad possible market: chipmakers, electronics companies, and organizations that need more specialized hardware could all have reason to care if the system works as promised.
Why chip design is so difficult
The reason this problem is valuable is also the reason it is hard. Computer chips contain millions to billions of logic gate components on a silicon wafer. Designers must decide where those components go while accounting for performance, power use, and other design needs.
In plain terms, chip design is not just drawing a map. It is a search through many possible arrangements, where a small change can affect whether the final chip meets its goals.
Alpha Chip approached the problem by using a reward signal. The design agent received a rating for how good a design was, then used that rating to update the parameters of its deep neural network. After thousands of designs, the founders say the agent became better and faster.
Goldie said Alpha Chip “could generate a very high-quality layout in, like, six hours. And the cool thing about this approach was that it actually learns from experience,”
Ricursive is trying to extend that idea. Goldie said the company’s AI chip designer will “learn across different chips,” meaning each project should improve the system for future chip projects.
The founders’ unusual overlap
Goldie and Mirhoseini’s partnership is central to Ricursive’s story. Their paths first crossed at Stanford, where Goldie earned her PhD as Mirhoseini taught computer science classes.
From there, their careers closely tracked each other. Goldie put it this way: “We started at Google Brain on the same day. We left Google Brain on the same day. We joined Anthropic on the same day. We left Anthropic on the same day. We rejoined Google on the same day, and then we left Google again on the same day. Then we started this company together on the same day,”
At Google, their collaboration was close enough to earn a nickname. Internally, they were called A&A. Their Alpha Chip work also had a more playful internal label from Jeff Dean, who called it “chip circuit training,” a reference to the circuit training workouts they both enjoyed.
The work brought recognition, but also controversy. In 2022, Wired reported that one of their Google colleagues 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 important AI chips.
What faster chip design could change
The Ricursive pitch is not only about speed for its own sake. Goldie described chips as core infrastructure for AI: “Chips are the fuel for AI,” she said. “I think by building more powerful chips, that’s the best way to advance that frontier.”
Mirhoseini argued that the length of the chip-design process is limiting the pace of AI progress. Her view is that faster design could support “this fast co-evolution of the models and the chips that basically power them,”
The founders also point to hardware efficiency as a nearer-term benefit. If AI labs can design more efficient chips and eventually more efficient underlying hardware, the source article says their growth would not have to consume so much of the world’s resources.
Goldie gave one example of what that could mean: “We could design a computer architecture that’s uniquely suited to that model, and we could achieve almost a 10x improvement in performance per total cost of ownership,”
Ricursive has not named its early customers. The founders say they have heard from every big chip making name you can imagine and have their pick of first development partners.
For now, the company’s central question is execution. The founders have already shown that AI can help produce chip layouts. Ricursive is betting that the same idea can become a broader platform for how chips are designed.