AMD crossed $1 trillion in market value for the first time on Monday, Sept. 21, as investors bought CPU makers on a bet that Meta’s Muse personal agent will require far more server processors. Arm shares climbed 17.16% that day as the wager spread across the sector. AMD became the fourth U.S. chipmaker to reach the threshold after Nvidia, Broadcom and Micron, and server CPU suppliers led the gains.

What Changed

AI-generated summary, reviewed by an editor. More on our AI guidelines.

The CPU argument

Meta launched Muse on Sept. 8 with a design that gives each user a dedicated virtual computer in the cloud. The agent can open a browser, fill out forms and keep working after a user closes the app. Persistent virtual machines, headless browsers, software calls and data transfers rely heavily on CPUs, while GPUs handle the intensive calculations behind AI models.

Estimates for agent workloads ranged from four CPUs per GPU to 40 CPUs per GPU. Intel Chief Executive Lip-Bu Tan said in recent days that his company could supply only half the CPU demand it was seeing from frontier-model companies. Arm, which rose to $322.90 at Monday’s close, struck a partnership with Meta earlier in 2026 to codevelop CPUs for Meta data centers. It entered the server CPU market in March 2026 with its first in-house chip for AI data centers.

The evidence for Muse itself is much thinner. Meta’s launch materials and Monday’s reporting did not specify how much CPU capacity Muse consumes or whose processors run it. The ratio estimates span a tenfold range, and the published estimates were not tied to measured Muse workloads in the reporting.

More than one catalyst

AMD closed up about 10% on Monday, while Intel gained 12.14% and had a market value of about $645 billion that day. The Philadelphia Semiconductor Index rose more than 3%. Meta, AMD’s second-biggest customer, accounted for about 5.5% of the chipmaker’s revenue.

Investors were also looking toward an expected Trump-Xi state dinner with AI executives, a possible SK Hynix deal with Intel, and Nvidia Chief Executive Jensen Huang’s forecast last week that his company would double chip shipments in 2027. Falling oil prices and Treasury yields supported the broader market on Monday.

The rally also followed a chip-stock slump from a July 2026 peak, when concern about excessive AI infrastructure spending had grown. Amazon blocked Muse from its store on Sunday, Sept. 20, a constraint on the agent’s reach. Monday’s reporting did not separate how much of the rally came from Muse and how much came from the other catalysts.

Know someone who'd find this useful? ✉️ Email it to a friend in one click, or they can subscribe free here.

AMD’s results and outlook

AMD entered Monday on a five-session advance of about 25%, lifting its gain since the start of 2026 above 180%. Its second-quarter revenue reached $11.54 billion, up 50% from $7.69 billion a year earlier. Data-center revenue rose 107% to $6.7 billion over the same period, though AMD shares fell after the August report because its forecast missed elevated expectations.

Chief Executive Lisa Su said in August that AMD expects to double data-center sales in 2027. Last week, BofA analyst Vivek Arya raised his estimate for the chip market in 2030 to $3.2 trillion from $2.7 trillion. Nvidia’s roughly $5.4 trillion value on Monday remained more than five times AMD’s new threshold.

If persistent personal agents become a large computing market, how quickly can CPU supply follow agent demand?

Frequently Asked Questions

Why did AMD, Intel and Arm stocks jump on Sept. 21?

Investors bought CPU makers on a bet that Meta's Muse personal agent will require far more server processors. AMD closed up about 10%, Intel gained 12.14% and Arm climbed 17.16%. Other catalysts included an expected Trump-Xi state dinner, a possible SK Hynix deal with Intel, and falling oil prices and Treasury yields.

Why would a personal AI agent need CPUs rather than GPUs?

Muse gives each user a dedicated virtual computer in the cloud that can open a browser, fill out forms and keep working after the app closes. Persistent virtual machines, headless browsers, software calls and data transfers rely heavily on CPUs, while GPUs handle the intensive calculations behind AI models.

How many CPUs could agent workloads need per GPU?

Estimates range from four CPUs per GPU to 40, a tenfold spread. The published estimates were not tied to measured Muse workloads, and Meta's launch materials did not specify how much CPU capacity Muse consumes or whose processors run it.

Is CPU supply keeping up with demand?

Intel Chief Executive Lip-Bu Tan said in recent days that his company could supply only half the CPU demand it was seeing from frontier-model companies.

How does AMD compare with Nvidia after the milestone?

Nvidia's roughly $5.4 trillion value on Monday was more than five times AMD's $1 trillion threshold. AMD's stock is up more than 180% since the start of 2026, and CEO Lisa Su expects to double data-center sales in 2027.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Meta Turns to Amazon Graviton for AI Agents
Meta's AWS Graviton deal puts CPUs back in the AI infrastructure story. GPUs still train the models, but agents need search, code execution, orchestration, and post-training work. That shifts the bottleneck to a quieter chip layer.
Implicator.ai
AMD Data Center Revenue Jumps 107% to $6.7 Billion in Q2 202
AMD's data center business more than doubled to $6.7 billion and the company guided above consensus for the third quarter. The stock fell about 9% anyway, with analysts pointing to a Helios forecast that fell short of expectations and capital spending that nearly tripled from a year earlier.
Implicator.ai
Meta Muse Tops US App Store Ahead of ChatGPT
Meta's Muse hit No. 1 on Apple's US free iPhone chart on September 18, ten days after launch, ahead of ChatGPT, Gemini and Claude. Sensor Tower download estimates trail Meta's own launches, and 8% of 1,500 US consumers surveyed said they would trust Meta with their passwords.
Implicator.ai
AI News

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: editor@implicator.ai