Here is the thing about AMD right now. The stock has had a remarkable run, analysts keep lifting their targets, and the Helios platform is generating genuine enterprise excitement. But ask most investors what is actually driving it, and the answer you get back is GPUs. Instinct accelerators. The Nvidia alternative trade.
That framing is not wrong. It is just incomplete.
The numbers from Q1 first. AMD reported Q1 2026 non-GAAP EPS of $1.37, while revenue came in at $10.3 billion, up 38% year over year. Free cash flow rose to $2.566 billion for the quarter. The stock surged on the results. Most coverage moved on quickly.
What got less attention was a single line in CEO Lisa Su’s commentary about CPUs.
The Forecast That Doubled Overnight
During the Q1 earnings call, Su said AMD now expects the server CPU total addressable market to grow at greater than 35% annually, reaching over $120 billion by 2030—up sharply from its prior roughly 18% CAGR outlook shared in November 2025 (which implied about $60 billion by 2030).
The reason is agentic AI. Not the chatbot kind. The kind where multiple AI models act autonomously, call tools, execute workflows, and hand off tasks to each other in real time. Instead of the previous 1:4-8 CPU-to-GPU ratio with chatbot AI, agentic AI is moving toward a higher CPU requirement, with more CPU compute needed for orchestration and infrastructure support. That is a structural shift in how data centers get built and how compute budgets get allocated.
As AMD CFO Jean Hu has emphasized in public remarks around earnings, agentic AI workloads increase demand for CPU compute for orchestration and broader infrastructure needs.
Slight tangent, but it matters: on the Q1 call, AMD said server CPU sales grew by over 50% year over year, with both cloud and enterprise customers up more than 50%.
What Is Launching This Week
AMD CTO Mark Papermaster has said EPYC Venice, the company’s next-generation (Zen 6) server CPU, will be officially introduced at AMD’s Advancing AI event on July 22 and 23.
EPYC Venice is tied to TSMC’s 2nm process: AMD has said Venice is ramping production in Taiwan on TSMC’s advanced 2nm process technology, with future plans to ramp production at TSMC’s Arizona fabrication facility.
Reporting ahead of the event indicates the flagship configuration goes up to 256 Zen 6 cores, with a new SP7 socket and 16-channel memory support delivering up to 1.6 TB/s of bandwidth, and PCIe Gen 6 support.
Venice is the host CPU inside the Helios rack-scale platform. Helios integrates 72 MI455X accelerators at rack scale, and AMD markets the rack at up to 2.9 exaflops of FP4 compute performance and 31 TB of HBM4 memory. AMD also specs MI455X at up to 432 GB of HBM4 memory and 19.6 TB/s of memory bandwidth per GPU.
Microsoft has announced it will deploy AMD’s Helios rackscale platform on Azure for production-scale AI inference, and has also described two new Azure VM series built on AMD’s upcoming EPYC Venice CPUs.
The Business Behind It
Data Center segment revenue climbed 57% to $5.8 billion in Q1, driven by AMD EPYC processors and Instinct GPUs, while Client and Gaming revenue rose 23% to $3.6 billion and Embedded grew 6% to $873 million.
Non-GAAP gross margin was 55%, up 170 basis points year over year. AMD’s Q2 2026 outlook points to revenue of about $11.2 billion (plus or minus $300 million) with non-GAAP gross margin around 56%. Q2 earnings are confirmed for August 4.
What Wall Street Is Saying
The analyst community has spent the past several weeks catching up to the CPU story. Goldman Sachs raised its price target to $640 in early July. Wells Fargo raised its target to $615 on June 30. Cantor Fitzgerald raised its target to $700, and UBS raised its target to $670 in late June.
Wells Fargo increased its server CPU revenue estimates for AMD, now modeling $16.0 billion for 2026.
Where the Risk Lives
Nvidia’s CUDA software ecosystem remains a real competitive advantage. AMD’s ROCm stack, while improving, does not yet match the breadth and depth of Nvidia’s CUDA ecosystem for the full range of AI and HPC workloads. Developer inertia is slow to reverse.
A meaningful caveat on supply: specific claims about HBM4 being “fully allocated through 2026” and MI455X not reaching mass production until Q2 2027 are difficult to verify from primary sources. Supply availability could still constrain ramp timing, but the precise allocation and mass-production timing should be treated as uncertain.
China remains an overhang. AMD’s filings discuss inventory and related charges associated with U.S. export controls on AMD Instinct MI308 Data Center GPU products. Further restrictions could weigh on addressable revenue. And the stock is not cheap. The valuation leaves little margin for an execution stumble.
The Three Dates That Matter Now
- July 22-23: AMD’s Advancing AI 2026 conference runs at the Moscone Center in San Francisco, with Lisa Su’s keynote on July 23 at 9:30 AM PT. Watch for full Venice SKU details, Helios customer disclosures, and any roadmap updates.
- August 4: Q2 earnings.
- H2 2026 Helios deployment execution: AMD has said Helios production shipments are on track to begin in the second half of 2026, and Microsoft has announced it will deploy Helios on Azure.
Bottom Line
The GPU trade at AMD is real. The MI450 pipeline, the Meta 6-gigawatt commitment, the Helios rack delivering 2.9 exaflops per unit. That story has earned its place in the conversation.
But the part of the AMD thesis that still does not get enough credit is this: AMD’s most important growth driver in the next two years may not be the Instinct GPU. It may be the EPYC server CPU. While AMD previously outlined a server CPU market growing at roughly 18% annually, AMD now expects the total addressable market for server CPUs to grow at greater than 35% annually, reaching over $120 billion by 2030.
The Advancing AI event this week and August 4 earnings will either deepen the conviction behind that call or start testing it. Either way, the CPU angle is worth understanding before the results land.
For informational purposes only.
