What it means
A multimodal model takes different types of input, like a picture and a written question, and reasons across them in a shared representation, then produces an answer that may itself span text, images, audio or video. It works by encoding each modality into a common internal space so the model can relate a caption to an image or a sound to a scene. In the AI supply chain these models sit in the workloads layer and broaden where AI can be applied, from document understanding to video generation. The trade-off is resource intensity: images, audio and video pack far more data per request than plain text, so multimodal workloads enlarge context, memory footprint and compute per query, intensifying pressure on accelerators and serving infrastructure.
Why it matters to investors
Multimodal models expand the addressable set of AI applications but intensify resource needs per request, which raises the compute and memory demand that flows down to accelerators and data centers. Builders such as Runway, MiniMax and ByteDance compete on generation quality while carrying heavier serving costs.
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 Multimodal model 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.