San Francisco — May 30, 2026: A new Wolfe Research analyst note has put Advanced Micro Devices (NASDAQ: AMD) firmly back in the spotlight, outlining a compelling multi-year growth thesis built on two powerful pillars: a ~30% expansion of the CPU Total Addressable Market by 2028 driven by agentic AI workloads, and a quietly growing 4% share of China's AI chip market through AMD's open-source ROCm software stack — even as U.S. export controls tighten the competitive landscape. AMD reached its all-time high on May 22, 2026 at $481.41 — with a current market capitalization of approximately $844.80 billion, up 11.05% over the last week alone.
The Wolfe Research Case: Agentic AI Expands CPU TAM by 30%
Wolfe Research says AMD could gain significantly from the next AI spending wave: agentic AI may boost the CPU Total Addressable Market by approximately 30% by 2028, raising server CPU demand — while wafer supply tightness at TSMC could also affect AMD's share allocation favorably.
This is a nuanced but important thesis. Traditional AI infrastructure spending has been dominated by GPU conversations — Nvidia's H100, H200, and Blackwell chips have commanded the bulk of analyst attention. But agentic AI — systems where AI models autonomously take actions, coordinate tasks, and run persistent workflows — is fundamentally more CPU-intensive than pure inference workloads. Each AI agent requires orchestration, memory management, and task scheduling that runs on CPUs, not just raw tensor operations on GPUs. As agentic AI deployments scale across hyperscalers and enterprise customers, server CPU demand expands dramatically — and AMD's EPYC processors are positioned to capture a substantial portion of that growth. For investors seeking deeper context on AMD's competitive roadmap, AMD's Investor Relations SEC filings provide authoritative access to the company's financial guidance and strategic disclosures.
China: 4% AI Chip Share Through ROCm — and Why It Matters
AMD has approximately 4% share of China's AI chip market, and China accounts for about 20% of AMD's total revenue. Despite export controls, AMD stays active in PCs, gaming, and select data-center segments in China. AMD is pushing its open-source ROCm stack in China as a direct CUDA rival — and while export controls and less mature developer tools limit broader adoption, AMD is positioning ROCm for wider local use.
The China story for AMD is a high-stakes balancing act. On one hand, the U.S. Commerce Department has progressively tightened export controls on advanced AI chips — limiting AMD's ability to sell its highest-performance Instinct GPU accelerators into China. On the other, AMD's Q4 2025 results included AMD Instinct MI308 GPU product revenue to China of approximately $390 million — demonstrating that compliant, export-control-cleared products continue to generate meaningful Chinese revenue. ROCm's open-source architecture is strategically important here: by building a developer ecosystem in China that runs on AMD hardware, AMD creates long-term switching costs and demand pull that can survive even the tightest export regimes.
AMD's Broader AI Roadmap: MI400, MI450, and the $1 Trillion Market
AMD CEO Lisa Su stated at AMD's Financial Analyst Day: "With the broadest portfolio of products and our deepening strategic partnerships, AMD is uniquely positioned to lead the next generation of high-performance and AI computing." AMD said it expects to lead the AI compute market — estimated at $1 trillion by 2030 with a 40% CAGR — and projects its AI data center revenue to grow at an 80% CAGR over the next 3–5 years.
AMD's MI350 Series is the fastest-ramping product in company history, already deployed at scale, while the MI450 Series ("Helios" systems) launches Q3 2026 with rack-scale performance leadership. The MI500 Series is planned for 2027. AMD has also won hyperscaler commitments at Meta and Microsoft — crucial validation that enterprise customers are building genuine multi-vendor AI infrastructure strategies rather than remaining purely Nvidia-dependent.
ROCm vs. CUDA: The Software Gap Is Closing
AMD's hardware roadmap is increasingly competitive — but analysts have long flagged its software ecosystem as the bottleneck. While AMD has approximately a 32% hardware advantage in some metrics, CUDA still delivers 10–30% better real-world performance on many workloads due to software maturity. That gap, however, is narrowing. AMD's investment in ROCm — and the strategic importance of an open-source CUDA alternative in markets like China where Nvidia's dominance creates political and supply-chain risk — means ROCm has both competitive and geopolitical momentum behind it.
AMD Stock and Valuation: All-Time High Territory
At the current range, AMD trades at roughly 9.8x 2026 consensus sales of approximately $46 billion — well below Nvidia's premium multiple but above what a pure cyclical semiconductor name would command. Consensus 2027 revenue sits near $68 billion. The base-case price target for AMD by year-end 2028 is $330–$400, with the bull case stretching toward $450 if AMD executes flawlessly and the AI spending cycle sustains momentum.
For investors, the Wolfe Research note crystallizes what the most bullish AMD thesis looks like in 2026: a company that wins not just on GPU compute but on the CPU infrastructure layer of agentic AI, builds a durable software moat through ROCm, and captures meaningful China revenue through a compliant product strategy — even as Nvidia faces increasingly complex geopolitical headwinds. The 30% CPU TAM expansion by 2028 is not a guarantee, but if agentic AI scales as the most optimistic forecasters project, AMD's EPYC-and-ROCm combination could prove to be one of the most strategically positioned semiconductor plays of the decade.