Home / Wiki / I / Inference (AI)

Inference (AI)

Technology and VendorsUpdated 17 Nov 2026
Share

Inference, in artificial intelligence, is the use of a trained model to produce outputs from new inputs: answering a question, recognising an image, scoring a transaction or generating text. It follows training, in which the model's parameters are learned. For a large language model, each response is generated piece by piece, and the volume processed is often measured in tokens.

A single inference request is far cheaper than training a model, but popular services handle enormous numbers of them, so total inference demand can exceed training demand over a model's life. It is served from AI data centers using GPUs or specialised chips optimised for speed and cost per query, and newer models that reason through multiple steps raise the compute used per answer. Investors watch inference because it links AI usage directly to recurring spending on chips, memory, power and cloud capacity.

Last updated 17 Nov 2026
Share this entry

Ref 1300000265