You're going to hear the term "AI PC" a lot throughout 2024. It's a natural evolution of the progress made with artificial intelligence, marked by the release of OpenAI's ChatGPT in 2023. But you might not hear a lot about Mac computers as AI PCs. That's partly because Apple has been relatively quiet with AI. While competitors have released many AI tools so far, like Microsoft's Copilot and Google's Bard, Apple doesn't have its standout AI feature yet.

That said, don't count out Macs in the race to create the best AI PC. Apple computers, and even its mobile devices, have had Neural Engines long before "NPU" became the latest computer-spec buzzword. They have the power, memory management, market proliferation, and — surprisingly — app support. If you have a great Mac from the last three or four years, you already have one of the best AI PCs you can buy. Here are four reasons why that's the case.

4 Powerful Neural Engines

They're just NPUs and are stunningly fast

Great AI performance is closely related to the quality of the Neural Processing Unit (NPU) on a computer's processor or system-on-a-chip (SoC). While Apple's fancy term for an NPU is the Neural Engine, they're mostly interchangeable monikers. However, the Neural Engines on Macs are getting more powerful each year. For example, Apple's new M3 family of chips has a 60% faster Neural Engine than the M1 family from three years ago. Compared to last-gen M2 chips, the Neural Engine is 15% faster. While Intel and AMD are just starting to bring NPUs mainstream, Apple has done so for a while.

Apple is also doing this with AI in mind, even if end users don't always notice. In its press release for the release of M3 SoCs, Apple specifically noted that AI and ML processes in apps like Topaz, Adobe Premiere, and Final Cut Pro will get faster. All this is to say that, contrary to popular belief, Apple does have extensive experience building AI PCs, and its flagship Neural Engines are a big reason why.

3 Better RAM management

AI tools require memory, and nothing handles it better than macOS

Mac computers excel at AI-based tasks for more than just their Neural Engines. AI processes are intensive in RAM and VRAM usage, an area where Apple Silicon Macs stand out. Even though Apple Silicon SoCs share unified memory between the CPU and GPU cores on the integrated chip, they still manage memory better than Windows PCs. In fact, I tested an M2 Mac Mini with 8GB RAM extensively and found that it somehow manages to provide a great experience in 2024.

That's not to say Macs with 8GB, or maybe even 16GB, will be blazing-fast at completing AI tasks. AI requires a lot of RAM, specifically VRAM, and that doesn't change just because you use macOS. However, the advantage that Macs have over Windows PCs is that they can better manage and allocate memory usage. For example, a Windows PC with 32GB RAM and only 4GB VRAM can't do much if an AI process requires more VRAM. By comparison, a Mac with 32GB total unified memory can give a lot more VRAM to GPU cores than the Windows PC could.

2 Support for third-party apps

Tools like Whisper and Stable Diffusion can run on macOS

You might be surprised to learn that the Mac also has support for plenty of on-device AI tools. Since the Neural Engine is secure and on-device, running large language models (LLMs) locally can be better for performance, cost, and privacy. You can use LM Studio on a Mac, which lets you use just about any LLM on-device. There's also Llama 2, which can run locally as well via a few different methods. Audio and video AI software apps, like Stable Diffusion and Whisper, can run on the Mac natively, too. Some software limitations you can occasionally encounter on the Mac aren't present with the best AI apps, so the Mac is just as capable as the latest AI PCs.

1 They're out in the wild

AI Macs aren't coming; they've been here for a while

By far, the biggest reason the Mac is the best AI PC is because people actually own them. Intel's Core Ultra and AMD's XDNA architecture are just starting to appear on new computers, and people don't upgrade their laptops and PCs that often. It could be years before the average Windows user buys a system with an NPU and, thus, an AI PC. If you look at the Mac instead, millions of people already have a Mac that qualifies as an AI PC. Macs have been shipping with Neural Engines since 2020, starting with the M1 chip. That means the entry-level Mac Mini and MacBook Air from 2020 — which are going on four years old — are AI PCs.

It's hard to pinpoint the exact number of Macs with Neural Engines out in the wild. However, looking at the Mac's sales figures for the last three years should give us a good indication of the level of market proliferation that Apple Silicon Macs have achieved: $29.4 billion in 2023, $40.2 billion in 2022, and $35.2 billion in 2021. Now, not all of these sales are Apple Silicon Macs. However, it's more than reasonable to assume that most of them are and have Neural Engines.

Two generations of MacBook Air and three generations of MacBook Pro have had Neural Engines and can handle AI tasks on-device. The same won't be true if you pull out a Windows laptop from two or three generations back. Apple's advantage is this: Mac users don't have to go out and buy an AI PC specifically to use AI.

Should you buy a Mac over an AI PC?

This doesn't necessarily mean that you should buy a Mac if you want an AI PC. New laptops, like the Acer Swift Go 14 and the MSI Prestige 16 AI Evo, have performed quite well in our testing as AI PCs. But it's clear that as AI PCs become more popular, people will realize that Macs have been AI PCs for a while. There's probably no need to upgrade if you have a Mac because you already have an AI PC.