Start with a number.
On September 21 local time, AMD closed up 9.95% at $615.52 a share, with its market cap crossing $1 trillion for the first time.
What does that mean? AMD became only the fourth U.S. chip company to enter the "trillion-dollar club," following Nvidia, Broadcom, and Micron.
Here's the more striking comparison. When Lisa Su took over AMD in October 2014, the company was losing money year after year. Measured from the low point in the summer of 2015, AMD's market value has multiplied more than 370 times.
!AMD CEO Lisa Su at a product launch
This rally wasn't driven by GPUs. It was an AI agent.
You'd think the AI chip story was already tired — it's basically Nvidia and GPUs. But the catalyst for AMD crossing a trillion was a little different.
What directly lit the fire was Muse, the consumer-facing AI agent Meta launched in early September.
Muse can carry out multi-step tasks on a user's behalf in the background: booking travel, managing schedules, shopping online. How big did it get? According to Sensor Tower, it hit 730,000 downloads in its first five days, beating ChatGPT, Claude, and Grok. By September 18 it had climbed to number one on the U.S. App Store's free chart, and it held the top spot for days.
Once Muse took off, investors ran the numbers: an AI agent isn't the same as an ordinary chatbot.
A chatbot answers one question at a time. But an agent is constantly calling models in the background, sequencing tasks, and moving back and forth between several pieces of software. That means the server side has to carry a heavy load of reasoning and task orchestration — and that work runs mainly on CPUs.
!AMD shares jumped nearly 10% the day its market cap crossed a trillion
The overlooked CPU moves back to center stage
For the past few years, AI money poured entirely into GPUs, because training large models is what hits graphics cards hardest. But the logic has shifted: once agents spread at scale, the part actually doing the work, running errands, and coordinating is the CPU.
Analysts at Jefferies put it directly — as AI agents spread among consumers, server CPUs stand to benefit clearly from higher reasoning, orchestration, and infrastructure workloads.
And in this market, there are really only two players who can compete — Intel and AMD.
What makes AMD unusual is that it has more than server CPUs; it also has GPUs and a full data-center product line. In other words, it can make money from all three waves: AI training, inference, and general-purpose computing. Meta is also AMD's second-largest customer, contributing roughly 5.5% of revenue. The bigger Muse gets, the more convinced investors become that AMD's server chips will be bought in volume.
The fundamentals hold up too. AMD's second-quarter revenue hit $11.54 billion, up 50% year over year, with its data-center business at $6.7 billion, up a dramatic 107%. Lisa Su has said the company plans to double data-center revenue by 2027.
Behind this is a bigger migration of money
Pull the camera back, and the real significance of AMD crossing a trillion isn't just that there's another giant.
It says one thing: the money from AI is spreading from a handful of GPU companies across the entire computing infrastructure.
Chip stocks took off together that day — Intel up 12.1%, Arm up 17.2%, the Philadelphia Semiconductor Index up for five straight days. Investors started repricing for the "age of agents," taking another look at CPUs, servers, and every resource that keeps agents running in the background.
Interestingly, two days later on the evening of September 23, AMD shares fell back more than 2%, with the market cap briefly slipping back below a trillion to around $993.6 billion. That back-and-forth tug-of-war shows the market is still fiercely debating just how much hardware demand AI can really pull in.
Here's the point for anyone running a business
After all of that, the signal for brands and companies is clear.
Capital is the most honest thing there is. When the smartest money in the world starts betting hard on "AI agents need massive background compute," it's essentially wagering on one thing: the things that will run businesses, make decisions, and connect users in the future are AI agents working around the clock, 24/7.
This isn't science fiction. It's happening right now — and it's already showing up in stock prices and earnings reports.
So for every brand that wants to survive in this era, the question becomes very direct: as users increasingly learn about you, compare you, and choose you through AI agents, are you ready to make those agents understand you, trust you, and want to recommend you?
That's what GEO solves. Not getting people to click into your website, but making the AI that decides on their behalf think of you first in its answer.
CPUs and chips are the utilities of the agent era. And GEO is the wire that gets your brand switched on in that era.