Ricursive Intelligence founders: AI-designed chips are key to solving the compute crisis

At TechCrunch Disrupt on October 14, 2026, Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini detailed their startup's efforts to build AI systems that can design chips, learning from each iteration to improve subsequent designs and accelerate the overall chip design process.

Goldie said that compute demand is expected to grow by at least 10,000x over the next five years, and argued that the industry should focus on making chips, bandwidth, and fab usage more efficient rather than building vastly more data centers. She described a future in which humans focus on ideas and creative thinking while AI handles execution across areas such as chip design and software development.

The founders cited AlphaChip, an AI system they previously developed, which produced an unexpected circular "donut-shaped" chip component layout that ultimately proved successful. Rather than partitioning chips into dozens of smaller blocks, they said their approach optimizes entire chips and can generate complex patterns such as wave-like designs.

Goldie added that if chip design could be completed in weeks, or even one week, custom chips for applications such as space, AR, VR, and biological computing would become financially viable and more energy-efficient.