Evercore positive on AMD and ARM citing market share report By Investing.com

Evercore reiterated Outperform ratings on AMD, Arm and Intel after Q1 2026 server CPU share data showed AMD up 220 bps quarter over quarter and Arm up 140 bps, while Intel fell from 59% to 55%. Arm said Mercury data undercounts its server CPU units by about 30%, implying a 2025 server CPU TAM of 30.4 million units with Intel at 59%, AMD at 24% and Arm at 17%. The firm’s CPU renaissance thesis remains positive for AMD and Arm, though it highlights tightening server CPU constraints for Intel.

Analysis

The market-share shift matters less as a one-quarter print than as evidence the server CPU market is still in a capacity-constrained transition, where incumbency is becoming a liability. If Intel is structurally supply-bound while AMD and Arm are taking share, the second-order effect is that cloud capex can increasingly bypass legacy x86 bottlenecks and be allocated to designs that optimize power efficiency and time-to-deploy rather than raw compatibility. That tends to favor the names with the cleanest roadmap and the fastest pass-through of wafer availability into shipment growth.

The most interesting implication is for the ecosystem, not just the three CPU vendors. A wider CPU TAM means more sockets, but also more accelerators, memory, networking, and advanced packaging demand, because server refresh cycles tend to pull adjacent content with them. That is constructive for the semiconductor supply chain broadly, but the share math suggests Intel’s weakness is likely to be felt in platform attach rates and customer design wins over the next 2-4 quarters, not just unit shipments today.

Consensus may be underestimating how much of this is a duration trade rather than a pure share trade. If investors accept a multi-year re-acceleration in server CPU units, AMD and Arm can compound into both share gains and market expansion, which is a better setup than either alone; meanwhile Intel’s issue is not merely share loss but the risk that customers build new procurement habits around non-Intel architectures before Intel can reassert manufacturing credibility. The contrarian risk is that the undercall on Arm may be partially offset by weaker monetization per socket and slower enterprise qualification, so the near-term revenue surprise could lag the unit-share narrative even if the strategic thesis stays intact.

Key Decisions for Investors

• Go long AMD vs INTC as a 3-6 month pair trade; AMD has the cleanest combination of share gain and TAM expansion, while INTC faces both competitive and execution risk. Use a tight stop if INTC regains sequencing momentum or if AMD guidance fails to convert units into revenue leverage.
• Add to ARM on pullbacks for a 6-12 month hold; the stock should benefit if the market starts pricing the server CPU market as structurally larger, but expect higher volatility because the valuation already discounts a good portion of the renaissance thesis.
• For a more convex expression, buy AMD Jan-2027 call spreads financed by selling shorter-dated upside in INTC or by reducing cash equity exposure; this gives exposure to a 12-18 month share and TAM rerating with defined downside.
• Overweight semi capex beneficiaries such as AVGO and MRVL versus INTC over the next 2 quarters; if CPU sockets expand, adjacent networking, interconnect, and custom silicon content should see better-than-linear attach growth.
• Fade any sharp INTC relief rallies into the next earnings window unless there is evidence of supply normalization; the risk/reward remains skewed against sustained multiple recovery until customers see consistent platform execution for multiple quarters.