← All questions

Understand it / 20 seconds

What exactly is a GPU?

A GPU is a chip built to perform many similar calculations in parallel. That makes it useful for the matrix operations at the heart of modern AI. One chip is only the beginning: large systems connect many GPUs through very fast networks.

02Show meFollow the connection
  1. 01Numbers
  2. 02Parallel math
  3. 03GPU
  4. 04Networked cluster
  5. 05Model output

Here's what matters

The answer changes when these conditions change.

  1. 01A GPU is not the AI model itself
  2. 02Memory and networking can limit performance
  3. 03More chips increase power and heat
  4. 04Software and utilization determine how efficiently hardware is used
03Prove itOpen the machinery

This explainer uses definitions, not a disputed numerical claim.

As the research library expands, project-specific evidence will appear here rather than being filled with placeholders.