At TechCrunch Disrupt 2026, Ricursive Intelligence co-founders Anna Goldie and Azalia Mirhoseini presented their vision for a future where artificial intelligence plays a central role in designing its own hardware.
What Happened
The session focused on the concept of 'recursive intelligence,' where AI systems are employed not just for general tasks but specifically to optimize the complex engineering of the chips that power them. Goldie and Mirhoseini detailed how their company, Ricursive Intelligence, is developing tools that allow AI to iterate on hardware designs, potentially accelerating the pace of innovation in semiconductor development.
Why It Matters
As the demand for compute resources grows, the bottleneck of manual chip design has become a critical constraint for the AI industry. If AI can effectively design better hardware, it could lead to a positive feedback loop where more efficient chips enable more powerful models, which in turn design even better chips. This approach challenges the traditional reliance on human-led EDA (Electronic Design Automation) workflows and suggests a new paradigm for hardware-software co-design.
The Bottom Line
Ricursive Intelligence’s presentation highlighted a strategic shift in how AI infrastructure is developed, moving toward systems that can self-optimize the physical substrate of their own intelligence.