Compute
In AI, compute means the computing power — chips, servers, electricity and time — used to train and run models. It is one of the biggest costs in AI and one of the main limits on who can build the largest systems.
In one line, for a 12-year-old
Compute is the amount of computer muscle an AI needs to learn and to answer.
An example
Canadian researchers apply to public programs for compute time because buying enough chips themselves would cost millions.
Why it matters to people
Access to compute decides whether universities, start-ups and smaller countries can take part in AI, or only a few large companies.