Key Notes
- Meta’s Muse runs tasks in a dedicated cloud environment, highlighting the CPU capacity needed alongside the GPUs that process AI models.
- AMD’s data-center revenue rose 107% to $6.7 billion in the second quarter, with growth driven by both EPYC processors and Instinct GPUs.
- Meta uses multiple CPU suppliers, while Arm and Nvidia are expanding their agent-focused offerings, keeping competition central to the investment outlook.
Meta Platforms (NASDAQ: META) is giving investors another reason to watch Advanced Micro Devices (NASDAQ: AMD), as the growing use of its Muse personal AI agent highlights demand for the server processors that keep autonomous software working in the background.
AMD shares had gained 32% over the past month, while Intel (NASDAQ: INTC) rose 21%, CNBC reported on October 6. Those are rolling monthly returns, rather than moves in Tuesday’s trading session.
Facebook’s parent introduced Muse on September 8. The current investment debate concerns the infrastructure needed to support its adoption, rather than a new product launch or an exclusive AMD supply agreement.
Personal Agents Need Computers That Keep Working
Muse goes beyond answering a question in a chat window. Meta says it can research information, organize projects, open a browser and carry out tasks on a user’s behalf, continuing to work after the person closes the app.
The service runs in a dedicated cloud environment called Muse Secure VM. Meta’s technical explanation describes an isolated Linux computer with a browser, storage, memory and CPU resources, where the agent can execute code and use connected services.
That architecture brings more of a conventional computer’s workload into an AI product. Processing the model’s responses is only part of the job: the system also has to run applications, handle files and coordinate actions across services.
AMD describes three CPU-dependent functions in agent workflows: operating the agents, feeding accelerators and supporting the enterprise services they access. Its EPYC portfolio is designed to address those different requirements, with configurations optimized for density, speed and efficiency.
The implication is a broader infrastructure opportunity. A useful agent can generate demand for CPUs as well as GPUs, with the hardware mix determined by how much work it performs, which applications it uses and how efficiently providers allocate resources.
MarketSpeaker’s earlier coverage examined the September rally in chip stocks as Muse gained attention. The latest discussion puts more emphasis on whether that enthusiasm can translate into sustained processor orders and earnings growth.
AMD’s Data-Center Growth Provides a Financial Foundation
The opportunity is arriving alongside substantial growth in AMD’s existing business. In its second-quarter results, the company reported data-center revenue of $6.7 billion, up 107% from a year earlier and representing 58% of total sales.
AMD attributed that increase to demand for both EPYC processors and Instinct GPUs. The segment combines those businesses, so its revenue cannot be treated as a measure of CPU sales alone or as revenue generated specifically by Muse.
Total quarterly revenue reached $11.5 billion, up 50%, while the company guided for roughly $13 billion in third-quarter revenue, plus or minus $300 million. That outlook, issued in August, remains a forecast rather than a reported result.
For investors, the distinction between shipped products and a potential market matters. Rising use of agents strengthens the case for additional computing capacity, but actual supplier revenue depends on procurement decisions, deployments and the prices customers are willing to pay.
Nor is Meta committed to a single CPU provider. A spokesperson told CNBC that its infrastructure is designed to work across different processors, giving the company flexibility when acquiring capacity. That limits the case for treating every new Muse user as an automatic AMD sale.
Competition Extends Beyond AMD and Intel
Arm Holdings (NASDAQ: ARM) is pursuing the same shift. In March, the company announced its AGI CPU, moving into production silicon products and naming Meta as its lead partner and co-developer.
Arm positioned the processor around the coordination work required in AI infrastructure. Meta’s participation gives the CPU opportunity a competitive dimension: growing demand does not necessarily mean that spending will remain concentrated among the incumbent x86 suppliers.
Nvidia (NASDAQ: NVDA) is also expanding its CPU offering. The company’s Vera processor targets workloads including agent execution, code running and data processing, complementing its accelerator systems. MarketSpeaker has covered Nvidia’s Vera strategy and its broader agent ambitions.
Meta Must Balance Adoption With Computing Costs
Meta has expanded Muse’s potential audience through its September 29 business offering. The agent can connect with Facebook Pages, Instagram business analytics and external tools, helping owners analyze performance, prepare campaigns and identify work requiring attention.
Meta says Muse remains free for most needs, with subscription plans for heavier use. It also requires approval before publishing, sending messages or spending money. Those controls matter because business tasks bring the agent into systems holding customer information and financial records.
The economics remain important for shareholders. Meta spent $31.08 billion on capital expenditures, including finance-lease principal payments, in the second quarter and projected $130 billion to $145 billion for 2026. Those companywide figures cover much more than Muse.
The next test is whether regular agent use produces enough customer value to support its operating costs. For chipmakers, the corresponding question is which processors win those workloads and how much additional demand becomes recurring revenue.
