The AI laptop is quickly becoming the default pitch from PC makers. New devices are being marketed around their ability to handle artificial intelligence tasks, promising better performance, privacy and battery life for employees using AI at work.
But most workers already access AI through cloud services on the laptops they have. That raises a more difficult question for IT buyers: does an AI laptop offer a clear reason to upgrade, or is the term being used to sell devices with higher specifications before companies have established a real need?
Ole Bülow, Director of Business Development at enterprise hardware and lifecycle provider Egiss, believes the answer depends on what an organisation means by “AI laptop.”
He argues the industry is currently grouping two very different kinds of device under the same label: laptops with a specialist chip that can support selected AI features, and more powerful machines configured to run larger AI models locally.
“There are different views of what an AI laptop is,” Bülow told UC Today.
“There is what I call the marketing view from Microsoft, which is the label that says this is an AI-ready laptop. And then there’s the new breed of laptops, which are high compute, with a lot of memory and fast disks and a lot of disk space, which allows you to run more AI locally.”
An AI Label Does Not Equal Local AI
The “AI-ready” label usually refers to a laptop with a neural processing unit, or NPU. This is a chip designed to handle certain AI tasks efficiently on the device itself.
For employees, that can support functions including background noise removal, camera effects, live captions, transcription, dictation and other features in meeting and productivity applications. Handling those jobs locally can reduce battery drain and avoid sending every task to a cloud service.
However, Bülow said that should not be confused with running a full-scale AI assistant or large language model directly on a laptop.
“An AI-ready laptop, based on Microsoft branding, is essentially that you have what’s called an NPU on the device, that allows you to run certain parts of an app’s AI on the device,” he said. “But you can’t really run the large language models on the device.”
That distinction matters because many of the AI tools employees use today do not rely on the extra AI hardware in their laptop. ChatGPT, Claude and many enterprise AI services run primarily in the cloud, with the work carried out in remote data centres rather than on the employee’s device.
Bülow said this makes some of the current marketing difficult to justify.
“Personally, I think it’s a bit of AI washing,” he said.
“Most enterprises these days use ChatGPT or Claude in the browser or an app on the device. That doesn’t need that NPU part of the device.”
Higher Specifications Change the Conversation
For Bülow, a device becomes genuinely AI-capable when it has enough memory, storage and processing power to run larger models locally.
That is a different requirement from simply buying a mainstream business laptop with an NPU. It may mean selecting a system with substantially more memory and storage than a typical employee device, and it is likely to come at a higher cost.
Research from Egiss found that 55.6 percent of the notebooks it supplied between June and August 2026 had at least 32GB of memory and an NPU.
While the figure represents the company's own supply data, and not the overall enterprise PC market, Bülow said it shows that customers are increasingly ordering devices with higher specifications.
But he drew a clear line between “AI-ready” and “AI-capable” devices.
“When we state AI-capable, we boil it down to the memory part and the disk part,” Bülow said. “When you’re looking at the large language models and running them locally, truly locally, then you’re looking at 64GB of RAM, one to two terabytes of disk space.”
