UBS flags these 2 stocks as key beneficiaries of standalone CPU

UBS raised its price targets on AMD to $670 from $455 and Arm to $470 from $260, citing accelerating demand for standalone CPU deployments in agentic AI infrastructure. The firm expects AMD CPU server revenue of $23 billion in 2027, $29 billion in 2028, and $50 billion in 2030, while estimating Arm internal CPU revenue near $14 billion by 2030. UBS remains constructive on both names, but it also pushed back against overly bullish $300 billion CPU TAM expectations, suggesting DRAM constraints could cap the market.

Analysis

The real read-through is not just that AMD and Arm get a valuation lift, but that the market is still underpricing the scarcity premium in the memory layer. If standalone AI racks become more CPU-intensive before they become architecture-agnostic, the bottleneck shifts to DRAM availability and cost per node, which means the most durable earnings power may accrue to memory suppliers rather than the CPU designers. That creates a subtle loser/winner split: Intel faces a double hit from share loss and a more supply-constrained roadmap, while AMD can win share without needing the total market narrative to be as large as the most bullish sell-side cases imply.

The second-order effect is that this is a capex allocation debate disguised as an architecture debate. Hyperscalers will likely continue diversifying across x86 and Arm to avoid single-vendor concentration, but if core counts and throughput remain the gating factor, near-term deployments should skew toward the incumbent performance stack even when long-run ARM penetration rises. That makes the next 2-4 quarters more relevant for AMD than the 2028-2030 horizon the note focuses on, because investors will trade the pace of rack deployments and backlog conversion before they fully capitalize the eventual TAM.

Contrarian angle: the market may be overestimating how quickly agentic AI demand translates into revenue. A large theoretical CPU TAM does not matter if memory, power delivery, and packaging constraints cap deployment density, which would compress the implied upside for both AMD and Arm. The setup is therefore bullish for AMD on relative share capture, but the broader sector call is less clean than it looks — the best risk-adjusted expression is likely long AMD versus a weaker execution rival, while keeping a close eye on DRAM pricing and data-center lead times as the earliest indicators that the thesis is either accelerating or stalling.

Key Decisions for Investors

• Long AMD over the next 1-3 months on any post-earnings weakness: the asymmetry is favorable if investors start marking up standalone rack share wins before 2027 revenue numbers become credible.
• Pair trade: long AMD / short INTC for a 3-6 month horizon. Intel carries both competitive-roadmap risk and higher downside if the market starts pricing in slower CPU share recovery in AI server deployments.
• Own MU as the cleaner second-order beneficiary if DRAM tightness persists into the next 2-4 quarters. This is the better expression if the thesis is that memory scarcity, not CPU TAM, becomes the binding constraint.
• Avoid chasing ARM as the primary long here unless you want a longer-duration call option. The near-term upside is capped by product-generation limits, so the risk/reward is less attractive than AMD for the next two reporting cycles.
• Use call spreads on AMD rather than outright stock if entering after a sharp move: the thesis is real, but valuation sensitivity is high and the market can easily front-run the 2027-2030 numbers.