AI EncyclopediaChips & deep techData centers & power
AI infrastructure
AI 인프라
A catch-all name for the chips, servers, data centers, electricity and networks needed to train and run AI.
Overview
Infrastructure means the facilities that life rests on, like roads and water supply. AI infrastructure is what AI runs on. It includes the AI chips that do the calculating, the servers that hold the chips, the data centers that house the servers, the electricity and cooling equipment that go into them, and the networks that connect them all.
Sometimes it also includes the cloud, which lets people rent these facilities. Even a great AI model cannot be offered to more people if there are not enough facilities to run it. So AI companies put as much effort into securing facilities as into developing models. The obstacles are that building takes a long time and electricity is hard to get.
An everyday example
You can have a delicious recipe (the AI model), but without a kitchen, gas, refrigerators and delivery routes, you cannot open a restaurant. That whole kitchen and its equipment is AI infrastructure.
Why it matters
"AI infrastructure investment" is in the news every day. This term helps you see why chip makers, power companies and construction companies all show up in AI stories.
Also calledAI computing infrastructure