# Graphics Processing Unit (GPU)

> A chip with thousands of small cores. It does the heavy math for AI.

**Field:** Infrastructure  
**Category:** Hardware  
**Also known as:** GPU, Graphics Card

## Think of it like

Think of it like a kitchen with a thousand cooks. Each one only peels, but they all peel at once.

## Definition

A graphics processing unit is a chip with thousands of small cores working at once. It was built to draw game graphics, where millions of pixels need the same kind of sum. AI training needs exactly that kind of work, so GPUs became the main chips for AI. A GPU has its own fast memory, called video memory. A big model is often trained on thousands of GPUs together.

## Example

Jack trains an image model on his laptop CPU, and one epoch takes 6 hours. On a GPU, the same epoch takes 10 minutes.

## Why it matters

GPUs made deep learning practical. How many of them exist sets how fast new AI can be built.

## Common mistake

Looking only at GPU speed. The amount of video memory often decides which models it can run at all.

## Three principles

- **Thousands of cores.** Small and simple.
- **All at once.** The same sum on lots of data.
- **AI's main engine.** Training and running models.

## Related terms

- [Central Processing Unit (CPU)](https://learn.benibix.com/term/infrastructure/central-processing-unit)
- [Video Memory (VRAM)](https://learn.benibix.com/term/infrastructure/video-memory)
- [AI Accelerator](https://learn.benibix.com/term/infrastructure/ai-accelerator)
- [Deep Learning](https://learn.benibix.com/term/ai/deep-learning)
- [PCI Express (PCIe)](https://learn.benibix.com/term/infrastructure/pci-express)
- [Distributed Training](https://learn.benibix.com/term/ai/distributed-training)
- [AI Compute](https://learn.benibix.com/term/ai/ai-compute)

## Told apart from

- [CPU vs GPU](https://learn.benibix.com/compare/cpu-vs-gpu): People often think a GPU is just a faster CPU. It is a different kind of chip. A CPU has a few strong cores that can do any job, one step after another. A GPU has thousands of small cores that all do the same sum at once. AI needs the second kind of work, so it runs far faster on a GPU.

## Taught in

- [CPU vs GPU vs TPU](https://learn.benibix.com/learn/infrastructure-hardware/cpu-vs-gpu-vs-tpu): Computer Hardware, 4/8

![Three chips drawn as grids. The CPU has a few large cores. The GPU has thousands of small cores. The AI accelerator is mostly one big block of units that multiply grids of numbers.](https://learn.benibix.com/diagrams/hw-processors.svg)

---

Source: https://learn.benibix.com/term/infrastructure/graphics-processing-unit
Author: Behnia Zarei (a developer in Vienna, Austria, who builds and teaches data, AI and business intelligence at BENIBIX)
Publisher: BENIBIX Learn (https://learn.benibix.com/)
Last updated: 2026-11-19
Licence: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Quote freely with attribution.
