Teraflops, Gigawatts, and You [public]

​AI companies are expected to spend over $500 billion on chips and data centers in 2026—roughly what the US spent in 35 years building the entire interstate highway system, adjusted for inflation.

​Justin Lebar (core contributor to Triton, LLVM member, CppCon speaker) will start from the specs of a single GPU and work his way up to the worldwide buildout—connecting dollars, teraflops, liters of water, and gigawatts along the way.

​After this talk you'll be able to:

• ​Contextualize "$500 billion in AI spending" and know how big that actually is

• ​Develop an informed opinion on data center electricity, water, and climate impact

• ​Name the supply chain players that matter—TSMC, Samsung, ASML—and why

• ​Understand why chip geopolitics keeps making the news

​No AI background needed. 45–60 min + Q&A.