Intermediate · Systems

Accelerator

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. Any specialized hardware designed to run AI workloads faster or more efficiently than a general CPU.

Technical Definition

Any specialized hardware designed to run AI workloads faster or more efficiently than a general CPU.

How it works

Includes GPUs, TPUs, NPUs, FPGAs, and chips from Cerebras, Graphcore, Groq, and others. Choice depends on workload (training vs inference), model size, latency targets, and budget. The accelerator landscape is one of the most dynamic areas in tech.

Related Concepts

• GPU — Graphics Processing Unit — a massively parallel processor that powers most modern AI workloads.
• TPU — Tensor Processing Unit — Google's custom-designed AI accelerator chip.
• CPU — Central Processing Unit — the general-purpose chip that runs most software.