Building an AI Observatory

Artificial intelligence is a powerful new instrument for industry and science—but our ability to innovate is often constrained by the availability of computing resources and the limitations of legacy, general-purpose chips. Advancing the frontiers of AI and science compels us to think differently about computing. At Cerebras Systems, Dr. Andy Hock and his colleagues saw this challenge as an opportunity and developed a new type of computer system, designed from the ground up for AI and powered by the largest processor ever built: a single wafer-scale chip.

These systems grew from first-principles thinking about AI computation and physics-informed innovations in materials and packaging—and today rank among the fastest AI systems in the world. This talk will explore some of the physics and engineering challenges of computing at this scale, why wafer-scale integration can transform the performance of AI systems, and what increasingly powerful AI may enable for industry, science, and human innovation.

Along the way, Andy will trace a path that began with physics, astronomy, and computing at Colgate and our Foggy Bottom Observatory—and ultimately led to building what he thinks of today as an AI observatory.