What it means
A context window is the maximum amount of text, measured in tokens, that a model can take in and consider at once, spanning the user's prompt, any retrieved documents, and the model's own response. It functions as the model's short-term working memory: anything outside the window is invisible to the model. Larger context windows let a model read whole documents, long conversations, or entire codebases at once, which enables use cases like document analysis and agents. But context is expensive, because attention cost and the memory needed to hold the key-value cache grow with length, tying model capability directly to concrete memory-capacity and bandwidth constraints on the serving hardware. In the AI stack, context length is a capability lever for applications and, at the same time, a driver of inference cost and hardware demand.
Why it matters to investors
Longer context windows unlock document-scale and agentic use cases, but they raise the memory and bandwidth each query consumes, linking model features directly to hardware demand. Labs that push long context, such as Moonshot AI and Google's Gemini, increase pressure on high-bandwidth memory and serving capacity.
Companies on this part of the chain
Named to show where the term sits in the AI supply chain — research, not advice, and never a recommendation to buy or sell.
Related terms
See Context window in the live AI chain.
THE ENTITY maps every constraint onto one live model — which part is tight now, who owns it, and who gets squeezed when it moves. Plain-English reads you can check.
THE ENTITY is an educational read on the AI supply chain — research, not investment advice. It explains how the chain works and who sits where, never price targets or buy/sell calls.