THE $200 BILLION CPU WAR
For years, investors believed AI was a GPU story.
Barclays believes the next chapter may be about CPUs.
The firm recently raised its estimate for the server CPU market to nearly $200 billion by 2030, driven by the rapid adoption of Agentic AI.
This is not a small revision.
It suggests that one of the most overlooked segments of AI infrastructure could become one of the largest opportunities in the semiconductor industry.
Why Agentic AI Changes Everything
Traditional AI workloads focused on training and inference.
Agentic AI introduces something different.
AI agents must plan tasks, manage memory, interact with external tools, coordinate workflows, and make multi-step decisions.
Those functions rely heavily on CPUs.
GPUs may perform the calculations, but CPUs increasingly serve as the orchestration layer of the entire system.
AMD Could Be the Biggest Winner
Barclays believes AMD is positioned to capture a significant share of future server CPU growth.
The company has spent years taking market share from Intel with its EPYC processors.
Now Agentic AI may accelerate that trend.
According to Barclays, AMD's CPU business alone could potentially double the company's earnings per share over time if current adoption trends continue.
Many investors view AMD primarily through the lens of its AI accelerators.
The larger opportunity may be hiding in plain sight.
NVIDIA Wants More Than GPUs
NVIDIA has no intention of remaining a GPU-only company.
Jensen Huang has steadily expanded NVIDIA's presence in CPUs through Grace and the newly announced Vera platform.
The company understands that future AI systems will require control, coordination, memory management, and networking—not just raw compute.
In modern AI servers, CPU and GPU value is increasingly approaching a 1:1 ratio.
That changes the economics of the entire industry.
Intel Faces Pressure From Both Sides
Intel once dominated the data center CPU market.
Today, the landscape looks very different.
AMD continues gaining market share.
NVIDIA is entering the CPU market.
Meanwhile, Intel is still working to restore its manufacturing leadership and improve competitiveness.
The company is no longer defending a monopoly.
It is fighting a two-front war.
The Real Opportunity Is the Entire AI Stack
Most investors are searching for the next NVIDIA.
But AI infrastructure extends far beyond GPUs.
The next decade may be defined by companies that control multiple layers of the compute stack:
• $Advanced Micro Devices(AMD)$ — CPUs
• $NVIDIA(NVDA)$ — GPUs + CPUs
• $Broadcom(AVGO)$ — Networking
• $Marvell Technology(MRVL)$ — Custom silicon
• $Micron Technology(MU)$ — HBM memory
• $Taiwan Semiconductor Manufacturing(TSM)$ — Advanced manufacturing
The biggest winners may not be individual chips.
They may be the companies building the entire ecosystem.
Final Thought
The last three years were dominated by GPUs.
The next three years may be defined by CPUs, memory, networking, and system architecture.
Barclays is making a bold bet:
The CPU is becoming strategically important again.
And the $200 billion CPU war is only getting started.
Post source X：@thequeennnz
Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.
