AI agents may be shifting some of the computing workload back towards CPUs, giving AMD and Intel a larger role alongside the GPUs that have dominated the AI boom. The change matters because agents need processors to keep their virtual computers and tasks running, not just chips to generate model responses.
AMD Watch analysis
What happened
CNBC reports that investors and analysts see growing use of personal AI agents as a source of demand for CPUs. Signal65 president Ryan Shrout said more agent use could shift workload from GPUs to CPUs. Futurum Group chief executive Daniel Newman described the split this way: CPUs perform agents’ workflows while GPUs handle the model’s thinking.
The report says Meta’s Muse and OpenAI’s Dots run in virtual machines, with benchmarks and the agents’ own responses indicating use of AMD EPYC processors. Meta told CNBC its system is designed to use whatever CPU is available, so AMD is not guaranteed the work. OpenAI also uses multiple CPU providers. Arm and NVIDIA are pursuing the market too.
Why it matters
An AI service’s hardware bill is not just a question of which chip runs the model. Agents that keep tasks going for hours or days also need computing capacity for their surrounding workflows. If that use grows, demand could spread across more of the server rather than concentrating so heavily on GPUs.
The report gives a reason AMD could benefit: its EPYC processors already have a foothold with cloud providers, and x86 compatibility helps run older software. But forecasts cited by CNBC are estimates, not confirmed orders, and Meta says it is CPU-agnostic. The agent boom may broaden the market without handing any one chipmaker a guaranteed prize.
Our read
This is a more interesting hardware story than “AI needs more chips”. Agents could change which chips do which jobs, and that creates an opening for established CPU makers as well as new challengers. AMD looks well positioned, but the practical test is whether reported interest turns into sustained deployments and sales. Nvidia is not obliged to sit this round out, either.
What to watch
- Whether cloud providers report rising CPU demand linked specifically to agent workloads.
- How much of that demand goes to AMD, Intel, Arm-based systems or NVIDIA’s CPU products.
- Whether agents’ operating costs and usage grow enough to make the shift commercially significant.
Discussion spark: If AI agents shift more work onto CPUs, will that meaningfully rebalance the chip market, or will GPU makers still capture most of the value?
Sources and evidence
- Meta Muse gives AMD a boost as AI momentum shifts to personal agents – CNBC (6 October 2026, 11:00 UTC)
Independent WittyWires tracker for public updates about AMD. Not affiliated with or endorsed by AMD; this is not an official account.