Ricursive Intelligence aims to slash chip design cycles from years to weeks

Designers Anna Goldie and Azalia Mirhoseini, the duo behind the startup Ricursive Intelligence, are set to take the stage at TechCrunch Disrupt 2026 to discuss how their systems learn across different projects to accelerate the creation of next-generation hardware. Before launching Ricursive, the pair co-led the AlphaChip project at Google, proving that artificial intelligence could generate chip layouts in mere hours—a task that previously consumed months of human engineering. Their early work proved instrumental in the development of multiple generations of Google’s Tensor Processing Units.

Ricursive Intelligence, which secured a $4 billion valuation and $335 million in funding within four months of its late 2025 launch, aims to push this concept further. By automating everything from component placement to design verification, the company seeks to reduce the traditional two-to-three-year chip development cycle to a matter of weeks. The firm’s backers include Nvidia, signaling strong industry interest in the potential for AI-driven hardware to catalyze future computational capabilities. Goldie, a Stanford computer science Ph.D., and Mirhoseini, an assistant professor at Stanford, are focusing on building a system that treats chip design as a cumulative learning process, where each new architecture informs and improves the next.