Skip to content
AI supply chain term

Tensor core

A tensor core is a specialized hardware unit inside a GPU that accelerates the matrix-multiply operations at the core of AI training and inference.

What it means

A tensor core is a dedicated arithmetic unit inside a modern GPU, built to perform matrix multiply-accumulate operations, the dominant math in neural networks, far faster than the GPU's general shader cores. Often implemented as a systolic array, tensor cores process small matrices in a single step and support reduced-precision formats such as FP16, BF16, and FP8, trading numerical precision for throughput and energy efficiency. They are the feature that turned graphics chips into AI accelerators, and their count and precision support largely define a GPU's advertised AI performance. In the supply chain the tensor core sits at the heart of the accelerator die; it is a lever because more capable tensor units raise compute per chip, but only if memory bandwidth and on-chip SRAM can keep them fed.

Why it matters to investors

Tensor cores are the differentiator that made GPUs the unit of AI compute, and their generation-to-generation gains drive the upgrade cycle buyers pay for. The merchant-GPU designer sets the pace the rest of the market follows.

Companies on this part of the chain

NVIDIANVDACoreWeave, Inc.CRWVNebius Group N.V.NBISMicrosoftMSFT

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

GPU / AI acceleratorGraphics Processing Unit — the massively-parallel chip that does most AI training and inference.SRAMSRAM is fast on-chip memory built from transistors that holds a processor's caches and register files, sitting physically next to the compute logic.FLOPs (compute throughput)Floating-point operations per second — a raw measure of compute throughput.Memory bandwidthHow fast data can move per second — often the binding limit, not raw compute.

See Tensor core 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.

Explore the features
All glossary terms

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.