Ricursive Intelligence is betting on AI-designed chips to tackle the compute crisis

Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini described at TechCrunch Disrupt an AI system that designs chips, learns from each iteration and improves subsequent designs. Goldie said demand for compute could grow at least 10,000 times over the next five years, possibly a conservative estimate, and argued that more efficient chips could reduce the need to expand data centers while making better use of bandwidth, wafers and fabs.

The founders say AI can discover unconventional architectures. AlphaChip, a system they previously developed, produced a circular arrangement of components described as donut-shaped that ultimately proved successful. Ricursive says it optimizes entire chips rather than splitting them into dozens of blocks, and that faster designs could make custom chips financially viable for applications including space, augmented reality, virtual reality and biological computing.