The mass market “isn’t quite ready” for AMD’s new class of PC, a leading analyst has asserted.
AMD on Friday laid out its vision for the ‘Agent Computer’, characterising it as “a new category of device built to run your AI agents full-time”.
“You do not operate it like a PC. You delegate to it,” the semiconductor giant explained in a blog post.
“A personal computer runs your apps. An Agent Computer runs your agents so they can run the apps for you. That is the shift.”
AMD claimed systems powered by its Ryzen AI Max+ processors, including the AMD Ryzen AI Max+ 395, are “exceptionally well suited to become Agent Computers”.
The trend is already “happening now”, AMD asserted, arguing that an Agent Computer is applicable to professionals, creators and developers.
“Specialised but fast-growing niche”
But is AMD’s vision more than just marketing claptrap?
Despite characterising it as a “highly compelling vision for the future of personal computing”, Omdia Research Manager Kieren Jessop claimed AMD’s Agent Computer “is a paradigm shift that the mass market isn’t quite ready for yet”.
The current memory crisis – which HP and HPE’s respective CEOs this month acknowledged will probably persist until 2027 – could also hamper uptake, Jessop added.
“Persistent shortages in high-bandwidth memory and other key components are driving price hikes that may keep these dedicated machines out of the reach of the average consumer for now,” Jessop told IT Channel Oxygen.
“There is also a significant technical barrier to entry; managing a two-device ecosystem requires a level of technical know-how that most users haven’t mastered.
“Consequently, we expect the concept of the ‘Agent Computer’ to remain a fast-growing but specialised niche, primarily adopted by developers, prosumers, and technical enterprise roles where the ROI of a 24/7 digital co-worker is most immediate and where the users understand how to extract that value.”
Marie-Christine Pygott, senior analyst at market watcher CONTEXT said AMD’s notion of the Agent PC takes AI PCs to a “higher level”, meanwhile.
“AMD says the full AI agent stack can be set up in under an hour. This signals the product targets the developer and enthusiast community – those looking to move away from cloud dependencies toward a self-contained, local AI environment,” she told IT Channel Oxygen.
“From what I can see regarding its offerings, AMD is effectively giving users a choice between high-capacity memory for simultaneous agents and high-velocity inference for speed-critical tasks. This makes the ‘Agent Computer’ concept highly specific compared to the rest of the market. I believe that the premium price is an issue when it comes to immediate Agent PC adoption, but AMD definitely provides a pretty concrete vision with this product of where high-end local AI computing is headed.”
Not every AI workload belongs in a hyperscaler’s datacentre, AMD stressed as it set out the rationale for the new classification.
“People and businesses want control over their data, affordable AI they can use every day without limits, and the confidence that their AI works for them. That makes local, privacy-centric, always-on agentic compute a real and growing need for consumers, creators, developers, startups, and SMEs,” it stated.
“That requires powerful hardware, high-bandwidth unified memory, efficient parallel compute, and the architecture to run sophisticated AI models.”
Jessop agreed that there is a “perfect storm” driving momentum behind AMD’s vision when it comes to increasing maturity in the software ecosystem and hardware capability.
This includes the February 2025 launch of Claude Code and November 2025 launch of OpenClaw, the latter of which he stressed already has over 300,000 stars on GitHub and sparked a run on Mac minis.
“This proves that the ‘Agent Computer’ category already exists in an ad-hoc form,” he concluded.
