My life & career
More efficient chips could eventually power better tools at lower energy use.
Imagine artificial intelligence helping engineers design a more efficient chip—then that chip helps run better artificial intelligence tools that assist the next design.
The computers behind artificial intelligence rely on some of the most complicated chips people have ever made. Designing those chips takes enormous expertise and involves exploring countless arrangements before a workable design reaches a factory. Now imagine artificial intelligence helping engineers search that design space more quickly, creating chips that could run future intelligence systems more efficiently. Improved hardware could then make some forms of research and design faster still. That is an intriguing feedback loop, though turning a clever chip design into something a factory can manufacture reliably remains a major challenge.
Big change is fascinating. Its implications are what matter.
More efficient chips could eventually power better tools at lower energy use.
Chip teams still need rigorous verification before algorithmic design suggestions are useful.
Semiconductors and electronics manufacturers confront rising complexity and capital requirements.
Jim’s electronics industry research explores design cycles and the expanding importance of edge AI.
Meet the futurist behind YottaBit ↗Which part of chip development is delayed by verification rather than idea generation?
Here's what researchers have demonstrated, what's still ahead, and where to check the source. It should deepen the story—not get in the way of understanding it.
What's happening today: TSMC confirms 2nm volume ramp; AI causality not established by that.
The next challenge: Design validation and the huge cost of making chips still set hard limits.
How the technologies connect: ML search + EDA + leading-node foundries.