CPU

Overview

The CPU (Central Processing Unit) is the processor found in every computer, built to execute a wide range of instructions one after another, quickly and flexibly. It is the default processor for general computing: running an operating system, managing files, executing logic. AI workloads have shifted the spotlight to GPUs, but the CPU still runs the system that everything else depends on.

How it works

A CPU has relatively few powerful cores, from a handful in laptops to more than a hundred in today's server processors, each capable of handling complex, varied instructions in sequence. This design makes it excellent at tasks with lots of branching logic and few repeated operations. GPUs took over AI workloads because they take the opposite approach: thousands of simpler cores built to do the same operation across huge amounts of data at once, which is exactly the kind of math AI training requires.

Why it matters

In an AI server, the CPU and GPU play different, complementary roles. The CPU manages the system, loads data, and feeds it to the GPUs, while the GPUs do the actual heavy computation. A poorly matched CPU can bottleneck an expensive GPU cluster by failing to feed it data fast enough, so choosing the right CPU is part of getting real performance out of GPU infrastructure. This is also why NVIDIA designed its own Vera CPU to pair with its Rubin GPUs.

Use cases