Faraj Aalaei Joins SemiAnalysis’s Dylan Patel at Lux Capital’s Future of Datacenters and Compute Symposium

On Wednesday, November 4, Cognichip CEO Faraj Aalaei joined SemiAnalysis’s Dylan Patel for a fireside chat at Lux Capital’s Future of Datacenters and Compute Symposium in Palo Alto, California. There, he dug into the immense impact physics-informed AI models are having on the chip design flow, and why the industry’s next big leap starts there.

Compute starts with the chip

Every conversation about the future of datacenters eventually comes back to silicon. The demand for new compute is growing far faster than the industry’s ability to design the chips behind it: a single advanced-node chip still takes years and tens of millions of dollars to bring to tape-out, and the pool of experienced designers is not growing to match.

Why physics-informed models

That gap is why Cognichip built Artificial Chip Intelligence ACI™ on physics-informed foundation models rather than general-purpose language models. Chip engineering demands rigor and mathematical precision, so ACI grounds its models in semiconductor physics and brings that intelligence to every stage of the flow, from specification and RTL through verification and power, performance and area optimization. The result is design work that once took months coming together in days.

Thank you to Lux Capital for hosting a terrific symposium, and to Dylan Patel for a sharp, wide-ranging conversation.