Infrastructure
AI Accelerator
ActiveUpdated: 28 August 2026Published
Key
innovation
Specialized hardware that accelerates AI computations (matrix multiply, convolution) with far higher energy efficiency than a CPU.
Category
Infrastructure
Abstraction level
System
Operation level
Architecture blockModel
Use cases
Training large models (LLMs, vision models)AI inference in data centers and at the edgeRobotics and autonomous vehiclesImage processing and computer vision
How it works
Accelerators combine many specialized compute units (e.g. Tensor Cores, systolic arrays) operating in parallel, support reduced precisions (FP16/BF16/FP8/FP4/INT8) and fast high-bandwidth memory (HBM) to maximize operations per watt and per second.
Problem solved
General-purpose CPUs are inefficient for the massively parallel tensor operations of neural networks, limiting speed and raising the energy cost of training and inference.