Fundamentals

GPU (Graphics Processing Unit)

A chip originally built to render graphics that turned out to be the ideal engine for AI, because it can do thousands of simple calculations at once instead of one at a time.

A CPU is built to do a handful of complex tasks in sequence, fast. A GPU is built the opposite way: thousands of small cores doing the same simple operation — mostly matrix multiplication — on different pieces of data simultaneously. That happens to be exactly the math a neural network is made of, which is why GPUs, not CPUs, became the default hardware for training and running AI models.

That accident of fit is also why one company's chips ended up so central to the entire industry, and why the debate over how defensible that position really is — whether it rests on superior silicon or on the software ecosystem built around it — has become a genuine antitrust question rather than just a market-share one.

Related analysis