# Graphics Processing Unit A GPU is a processor with thousands of small cores designed to do the same simple operation across enormous batches of data simultaneously. It was built to render graphics — calculating millions of pixels at once for video games and 3-D — but it turned out that the math behind training and running AI models ([[Foundation Model]]s, [[Frontier Model]]s) is the same kind of massively parallel arithmetic. So the gaming chip became the engine of the AI revolution. Where a [[Central Processing Unit]] does a few complex things fast, a GPU does billions of simple things at once. ## Why it matters here The GPU is arguably the most politically consequential consumer-electronics component of the decade. A few threads the vault tracks all run through it: The AI boom is a GPU boom. [[NVIDIA]] designs the dominant AI GPUs and rode them to become one of the most valuable companies on Earth, displacing older giants like [[Intel Corporation]]. "Compute" — the scarce input everyone in the [[AI Arms Race]] is racing to accumulate — is measured in GPUs, and stockpiling them is now a corporate and national strategy ([[Compute Sovereignty]]). [[Elon Musk]]'s xAI and the other frontier labs compete partly on who can amass the largest GPU clusters. GPUs are also the unit of account in [[Chip Export Controls]]. US restrictions on selling China advanced chips are written, in effect, to keep the most powerful GPUs out of Chinese AI and military programs — with NVIDIA repeatedly designing slightly slowed-down chips to thread the regulatory line. The GPU is where AI hype, semiconductor geopolitics, and the US–China contest converge into one scarce rectangle of silicon. ## See Also - [[Central Processing Unit]] — the general-purpose processor the GPU eclipsed for AI - [[Chip Export Controls]] — the policy regime built around GPU access - [[AI Arms Race]] · [[Compute Sovereignty]] — why GPUs are stockpiled like strategic reserves - [[Semiconductor]] — the broader chokepoint the GPU sits inside