Is AMD the hottest player in the next wave of AI narratives? Cloud-based agents (like Meta's Muse and similar products) are rapidly changing the CPU-to-GPU ratio. Traditional chat-based inference relies heavily on GPUs, but agents need to repeatedly perform tasks like tool calls, code execution, API requests, retrieval, and validation—most of which fall on CPUs. Microsoft Azure's measurements show that in some requests, the time spent on tool execution is already comparable to model inference time. Arm estimates that in the era of agents, the number of CPU cores required per gigawatt in data centers could increase from around 30 million to 120 million—roughly a 4x jump. AMD's CFO has also identified agentic AI server racks as the fastest-growing segment for CPU demand and has raised the total addressable market (TAM) for server CPUs to exceed $120 billion by 2030 (with some estimates even going beyond $200 billion), far higher than the previous expectation of around $60 billion. Channel sources and Morgan Stanley's estimates further support this: AMD's 6th-gen EPYC Venice chips are reportedly fully booked for 2027 production, with orders already being placed for 2028. Shipments in 2027 could rise from around 1.25 million units in 2026 to 6.75 million units (a 5.4x increase). This single product line, with higher average selling prices, could potentially generate around $51 billion in revenue visibility. AMD's current share of data center CPU revenue is already close to 46%. If the adoption of agents continues at this pace and AMD doesn't lose significant market share, the CPU business alone has the potential for multi-fold growth. Overall, the CPU demand driven by agents is already reflected in order books and TAM upgrades. Combined with the GPU business, analysts' consensus has raised AMD's revenue projections from around $35 billion in 2025 to approximately $160 billion by 2029 (a 4–5x increase). With more aggressive assumptions about agent adoption and market share, the combined growth of CPUs and GPUs could potentially match or even exceed this trajectory. #AMD #AI #Tech #DataCenter #CPUs #GPUs #AIRevolution #Investing #TechStocks