Bernstein raises INTC and AMD target prices but stays neutral

Overview

This is Bernstein's Q425 earnings preview for Intel and AMD. The report focuses on PC shipments, the server CPU cycle, progress on Intel 18A and foundry, AMD EPYC growth, and AI accelerator Helios/OpenAI orders, and adjusts both companies' 2026-2027 forecasts and target prices accordingly. The overall conclusion is neutral: fundamentals are improving in some areas, but valuation, share, yield, customer concentration, and market expectations still limit the upside view.

Core views

For Intel, the report raises its server market assumptions and expects 2026 Xeon revenue to grow by more than 20% year over year, but cuts its 2026 PC assumptions, arguing that after a strong 2025 there is a high year-over-year base, memory price increases could suppress shipments, and there is a risk of PC CPU oversupply. Although the 18A launch, foundry ambitions, and U.S. government support have improved market sentiment, the report still worries about continued client and server share losses, the need for better 18A yields, higher capital spending for foundry expansion, and minority interest expenses related to SCIP transactions that the market may be underestimating. For AMD, the report believes EPYC server share gains and the OpenAI agreement support the AI story, but Helios currently lacks more large announced customers, and its competitiveness versus NVDA Rubin remains questioned in terms of power consumption, size, performance, and software ecosystem. Both companies keep a Market-Perform rating.

Analysis framework

The report uses an earnings-preview and model-revision framework, mapping industry-level PC shipments, server demand, AI accelerator orders, and the competitive landscape to company revenue, EPS, target price, and ratings. The analytical focus is on comparing Bernstein's own forecasts with market consensus and identifying the business assumptions driving forecast divergence, including the strength of the server recovery, PC demand pressure, the AI revenue slope, foundry yield and capex, and the effect of accounting items on profitability.

Methodology notes

• Earnings forecast revisionRevenue/EPS model comparison

  Compare company Q425, Q126, 2026, and 2027 revenue and EPS with prior forecasts and market consensus.

  The report updates INTC and AMD models by raising server assumptions and lowering PC assumptions, while emphasizing that both companies' forward estimates remain below consensus.

• Industry demand analysisPC and server cycle assessment

  Use IDC/Gartner PC shipments, Taiwan ODM notebook shipments, and Xeon/EPYC growth rates to judge end-market demand.

  PC shipments still grew about 9-10% year over year in Q4, but the sequential trend and seasonality were not strong; server demand is viewed as better supported.

• Valuation and narrative validationAI and foundry narrative stress test

  Assess whether Intel 18A/foundry and AMD Helios/OpenAI narratives can support current valuations.

  The report argues that Intel's improved narrative has not resolved share, yield, and capex issues, while AMD's AI opportunity still depends on OpenAI progress and new customer announcements.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• US.INTC

  research target

  Strengths

  Improving server demand, the 18A launch improves the technology narrative, and U.S. government support provides a buffer in adverse scenarios.

  Weaknesses

  Client and server product share may continue to erode, PC CPUs may remain oversupplied for a long time, 18A yields still need improvement, and the foundry strategy may require higher capex.

  Comparison

  The report believes Intel's forward revenue and EPS remain below market consensus, and its fundamentals and valuation are not sufficient to justify a more positive rating.

  Risks

  SCIP-related minority interest expenses may be underestimated by the market, and PC demand and memory price pressure could affect 2026 shipments.

• US.AMD

  research target

  Strengths

  EPYC server share gains, the OpenAI agreement creates AI revenue ramp opportunities, and the core business should benefit from stronger server demand and share gains.

  Weaknesses

  Helios currently lacks other large announced customers beyond OpenAI, the AI narrative depends on order progress and new customers, and valuation already reflects a high degree of success.

  Comparison

  The report says AMD trades at around 35x P/FE, about 10 turns above NVDA, but certainty around AI customers and the ecosystem is still weaker than the market's optimistic expectations.

  Risks

  The OpenAI ramp pace, Helios competitiveness, AI gross margin, China/MI308 shipments, foundry/CoWoS supply, and PC demand are all key uncertainties.

Key data

• INTC Q425 forecast$13.5B / $0.10Above the prior $13.3B / $0.08 and slightly above consensus at $13.4B / $0.09.
• INTC 2026 forecast$52.2B / $0.47Above the prior $51.9B / $0.41, but below consensus at $54.1B / $0.62.
• INTC 2027 forecast$56.4B / $0.68Above the prior $55.9B / $0.64, but below consensus at $56.7B / $1.09.
• AMD Q425 forecast$9.7B / $1.31Slightly above the prior $9.6B / $1.30 and broadly in line with consensus at $9.6B / $1.32.
• AMD 2026 forecast$40.6B / $5.36Above the prior $40.3B / $5.32, but below consensus at $45.1B / $6.58; AI revenue forecast stays at $12B.
• AMD 2027 forecast$56.6B / $8.79Above the prior $55.5B / $8.58, but below consensus at $63.4B / $10.29; AI revenue forecast stays at $25.3B.
• PC shipmentsQ4 year-over-year growth of about +9% to +10%IDC/Gartner data show PC shipments are still strong year over year, but sequentially roughly flat and slightly below pre-pandemic seasonality.
• Valuation noteAMD around 35x P/FEThe report says this valuation is roughly 10 turns above NVDA, already embedding a fairly high success expectation.

Impact & implications

The report's investment takeaway is that small pre-earnings model upgrades may ease some concerns in the short term, but they are not enough to change the ratings. Intel's upside narrative comes from the server recovery, the 18A launch, and policy support, but investors still need to validate share, yield, foundry capex, and accounting expenses. AMD's upside comes from server share gains and the ramp of OpenAI AI orders, but more large customers and better AI gross margin are needed to support the high valuation. For the semiconductor sector, servers and AI remain the stronger themes, while PC may become a drag in 2026.

Risks

• Intel PC CPUs may be oversupplied; if 2026 PC shipments are hurt by memory price increases or weaker demand, revenue and profit could come under pressure.
• Intel 18A yields and foundry customer expansion are not fully proven; if deeper foundry investment requires more capex, cash flow and margins could be affected.
• Intel's share losses in client and server products may continue, offsetting the industry beta from the server recovery.
• If SCIP-related minority interest expenses are higher than assumed in market models, EPS could be pressured.
• AMD's AI narrative currently depends heavily on the progress of the OpenAI agreement; if there are not more large Helios customers, valuation support could weaken.
• AMD Helios' competitiveness versus NVDA Rubin remains uncertain in performance, power consumption, size, and software ecosystem.
• AMD AI revenue growth may come with gross margin pressure, especially when HBM content is high and supply chain costs are elevated.
• Section 232 tariffs, the China market, MI308 shipments, and supply chain capacity could affect revenue and profit for both companies.

What to watch

• Intel's comments on PC demand, channel inventory, the impact of memory price increases, and 2026 PC shipments in the Q425 call.
• Intel 18A yields, 18AP/14A, foundry customer engagement, EMIB/packaging capabilities, and related capex plans.
• Changes in Intel data center CPU share, Granite and Clearwater progress, and server cycle strength.
• The actual accounting treatment of SCIP-related minority interest expenses in 2026 and 2027.
• AMD's OpenAI business ramp, the GW scale it can deliver by 2027, and the $B/GW revenue assumption.
• Whether AMD announces more Helios large customers and the competitive feedback for Helios versus NVDA Rubin.
• AMD AI gross margin, HBM costs, foundry/CoWoS supply, and the overall gross margin trajectory.
• Whether AMD EPYC growth comes from true share gains or a cyclical server recovery, and whether PC share gains reflect pull-forward demand or channel fill.