Citi boosts chip sector outlook as semiconductor stocks rally

Citi projected the server CPU total addressable market will reach $237 billion by 2030, implying a 52% CAGR from $29 billion in 2025 and exceeding estimates from AMD, Arm and Nvidia. The firm identified AMD as the primary beneficiary of a CPU renaissance, with Intel a secondary beneficiary; Intel rose 9.8%, AMD gained 6.6% and Arm advanced 4.7%. The bullish chip outlook also lifted optical-equipment names, including Lumentum (+12.6%), Coherent (+8.8%) and Applied Optoelectronics (about +8%).

Analysis

The market is likely to reward the highest-confidence AI compute beneficiaries first, but the more durable implication is a mix shift from accelerator-only spending toward balanced server architectures. AMD has the clearest incremental earnings torque if enterprise and hyperscaler CPU refresh cycles broaden: CPU share gains improve gross margin and reinforce its ability to bundle EPYC with Instinct. Intel's rally is more fragile because CPU demand recovery does not solve its manufacturing-utilization, foundry-loss, and execution issues; absent evidence of sustained server-share stabilization, INTC remains a tactical catch-up trade rather than a structural long.

Optical names provide a cleaner second-order expression of rising cluster scale. COHR, LITE, and AAOI benefit only when announced AI infrastructure demand converts into deployments requiring higher-speed interconnects; this typically lags processor order visibility by one to two quarters. LITE has better near-term operating leverage but also greater valuation sensitivity after sharp momentum moves, while COHR offers a more diversified exposure to datacenter optics recovery. NVDA is not necessarily impaired by a CPU renaissance: more CPUs per AI cluster can expand total rack value, though a shift in budget allocation toward general-purpose compute could modestly dilute its accelerator-led multiple premium.

Consensus risk is extrapolating a long-dated market-size estimate into near-term revenue. The next 1-3 months require corroboration through hyperscaler capex commentary, AMD server revenue/gross-margin guidance, and optical inventory normalization; without these, recent gains can reverse as positioning unwinds. Over 6-18 months, the key falsifier for the AMD thesis is server CPU share failing to advance despite an expanding market, while Intel needs data-center revenue growth and gross-margin recovery—not analyst optimism—to justify a rerating.

Key Decisions for Investors

• Initiate a 3-6 month long AMD / short INTC pair, sized beta-neutral: AMD has superior share-gain and margin-conversion exposure, while INTC remains exposed to execution and fixed-cost deleveraging. Target 15-20% relative upside; exit if AMD data-center growth or gross-margin guidance is cut, or if Intel demonstrates two consecutive quarters of server-share stabilization with improving gross margin.
• Add COHR on pullbacks for a 6-12 month AI-interconnect deployment cycle; use a 7-10% stop from entry given cyclical optics volatility. Prefer COHR to LITE after momentum-driven upside because diversification reduces dependence on a single datacenter demand ramp.
• Do not chase AAOI or LITE immediately after high-beta moves. Set an entry alert around next earnings for confirmed 800G/1.6T order visibility, backlog conversion, and gross-margin expansion; absent those metrics, these are flow-sensitive trades rather than investable fundamentals.
• Maintain NVDA exposure rather than rotate wholesale into CPUs, but hedge near-term relative-risk with a modest long AMD / short NVDA overlay only if AMD reports material server-share gains. The trade fails if hyperscaler spending remains accelerator-concentrated and AMD's CPU momentum does not translate into revenue revisions.