The rise of AI PCs: A new endpoint MSPs will soon encounter

Managed service providers (MSPs) in the age of artificial intelligence (AI) will soon see a different type of endpoint, known as AI PCs, on the networks they manage.

AI PCs simplify the process of building AI models and running inference engines, making them an attractive addition for organizations. Unlike traditional PCs, AI PCs include a neural processing unit (NPU), which enables them to perform tasks such as learning, reasoning, and problem-solving without relying on cloud services.

The NPU provides dedicated computer resources that allow a PC to run these applications locally, eliminating the latency that would otherwise occur when accessing a cloud service over a wide area network.

Growing adoption and competing NPUs

In general, AI PCs come in two flavors. There are workstations that are typically used by data scientists to create AI models and the PCs that are used to optimize the running on the inference engines that AI applications depend on. Currently, Intel, AMD and Qualcomm have each developed competing NPUs, while Apple has developed its own silicon.

It’s still early days so far as adoption of AI PCs is concerned, but a survey of over 850 IT leaders conducted by The Futurum Group finds 62 percent plan to allocate budget dollars this year to the acquisition of AI PCs, with more than half (55 percent) citing a need to make their workforce more competitive as quickly as possible.

The survey notes that just over half of respondents expect to use NPUs provided by Intel. Nineteen percent of respondents expect to use AMD, while 15 percent expect to use Qualcomm. Apple ranks fourth, with 11 percent of respondents selecting M3 and M4 processors as their top choice. These processors provide access to an NPU embedded in a system-on-chip (SoC).

Unlocking potential with smart security

AI PCs are naturally more expensive than traditional PCs. Over two-thirds of IT leaders surveyed (68 percent) are concerned about a value gap due to the current lack of AI applications. As more of these applications become available, that issue will start to dissipate. In fact, according to the survey, only 36 percent of respondents expressed concern about the higher cost of AI PCs.

Of course, as soon as an organization deploys an AI PC, cybercriminals will inevitably take notice. Given the nature of the applications being deployed on these platforms, they immediately present a rich target that will be too tempting for cybercriminals to ignore.

Managing AI PCs should be straightforward, with minimal challenges beyond occasional updates to the AI models. This should not add significant complexity to existing workflows. Still, the one thing that is certain is that cybercriminals will be looking to exploit any and all AI weaknesses they can discover.

Photo: artin1 / Shutterstock