Ricursive Intelligence, a startup focused on building an AI system that designs and automatically improves AI chips, has raised $300 million at a $4 billion valuation, marking a major early-stage funding milestone. The company confirmed that Lightspeed led the round, underscoring strong investor confidence in its ambitious vision.
According to the company, Ricursive is developing a system that can create its own silicon substrate layer and rapidly accelerate AI chip improvements. Consequently, the founders believe this continuous, self-improving loop could eventually push the system toward artificial general intelligence (AGI).
Notably, this Series A round closed just two months after Ricursive formally launched, following a seed investment led by Sequoia. As a result, the startup has now raised $335 million in total funding, according to reports.
Ricursive Anna Goldie (CEO) and Azalia Mirhoseini (CTO), both former Google researchers, founded the company to commercialize their cutting-edge research. Importantly, their work on a novel reinforcement learning method for designing chip layouts—known as AlphaChip—has already powered four generations of Google’s TPU chips, according to the startup.
In addition to Lightspeed, the funding round also attracted prominent investors such as DST Global, Nvidia’s venture capital arm NVentures, Felicis Ventures, 49 Palms Ventures, and Radical AI, further strengthening Ricursive’s strategic backing.
Meanwhile, these two companies are not alone in pursuing this emerging category. As previously reported, Naveen Rao’s AI hardware startup, Unconventional AI, is also developing an intelligent substrate. In December, the company raised a $475 million seed round at a $4.5 billion valuation, with Andreessen Horowitz and Lightspeed Ventures leading the round, alongside Lux Capital and DCVC.
Ricursive Intelligence’s rapid fundraising and bold technological roadmap highlight growing investor conviction in AI systems that design and improve their own hardware. As multiple well-funded startups converge on this self-improving AI paradigm, the race to redefine chip design and push the boundaries of AI capability is clearly accelerating.
