Artificial intelligence requires increasingly powerful chips to run its complex workloads. However, designing those advanced microchips remains an exceptionally slow, intricate, and labor-intensive process. This raises a fundamental question of what happens when machine learning begins to shoulder some of that heavy engineering work itself.
Accelerating the Chip Design Cycle
That is the core premise behind Ricursive Intelligence. The startup is developing specialized artificial intelligence capable of designing silicon chips, learning continuously from the engineering process, and improving subsequent iterations. If successful, this approach could drastically accelerate the development of the underlying hardware that powers modern computing.
Developing a commercial semiconductor typically requires anywhere from two to three years of meticulous engineering. Ricursive aims to compress that entire development cycle down to just a matter of weeks, fundamentally changing how hardware ecosystems evolve.
Proven Expertise in Silicon Engineering
Anna Goldie and Azalia Mirhoseini have already demonstrated that artificial intelligence can drive meaningful breakthroughs in chip design. Before launching Ricursive Intelligence, the pair co-led AlphaChip at Google, an AI framework capable of generating complex chip floorplans in mere hours instead of the weeks required by human teams. Their pioneering work contributed directly to designing multiple generations of Google Tensor Processing Units.
In addition to their work on AlphaChip, both researchers co-founded Google's Machine Learning for Systems group. They also served as early employees at Anthropic and held senior research scientist roles at Google DeepMind.
Academic Background and Rapid Funding
Goldie serves as the founder and chief executive officer of Ricursive, holding a computer science PhD from Stanford University and recognition as one of MIT Technology Review's innovators under 35. Mirhoseini serves as the founder and chief technology officer, holding an assistant professorship in computer science at Stanford alongside founding its Scaling Intelligence Lab.
The duo officially launched the company in late 2025, attracting immediate interest from major venture capitalists. Within a span of four months, the enterprise successfully secured $335 million in financing at a $4 billion valuation, which included a substantial $300 million Series A round featuring prominent backers such as Nvidia.
Creating a Self-Improving Hardware Loop
Ricursive intends to automate a much larger portion of the traditional chip development pipeline, stretching from initial component placement all the way through rigorous design verification. The system is designed to generalize across different semiconductor architectures, meaning insights gathered from one project directly inform and optimize the next.
This creates a powerful feedback loop where advanced AI assists in engineering superior hardware, and that superior hardware subsequently enables the creation of even more capable intelligence models. By bridging the developmental gap between software algorithms and physical silicon, the founders hope to unlock entirely new chip paradigms.
