AI Demand Combined with “Chip Inflation” Extends Enterprise Infrastructure Momentum, but Tactical Participation Is Still Needed

Overview

The report revises the previous view that memory inflation would suppress enterprise hardware demand. Enterprise customers increasingly see tight DRAM and NAND supply as a multi-year structural issue, so they are no longer waiting for prices to fall, but are instead pulling forward purchases of personal computers, servers, and storage equipment. AI capacity expansion represents a second demand tailwind, making the infrastructure cycle stronger and more extended than expected. However, current growth still mainly comes from refresh cycles, budget pull-forward, and infrastructure modernization, rather than a broad and durable enterprise AI supercycle.

Core views

Morgan Stanley upgrades its view on the North America IT Hardware industry to In-Line with the market and adjusts its global end-market preference to storage over servers, and servers over personal computers. The storage cycle typically lags servers by four to five quarters, has longer demand durability, and still has room for earnings forecast upgrades; server growth is the strongest, but the degree of pull-forward purchasing is also high; personal computer demand has already shown signs of weakening. Given that hardware stocks have risen by more than 100% since the beginning of 2025, and valuations of traditional enterprise hardware are also above historical peaks, the investment strategy should focus on high-quality names that still have room for earnings upgrades, margin improvement, and valuation re-rating.

Analysis framework

The report combines the AlphaWise enterprise hardware spending survey, CIO survey, checks with enterprise channels and value-added resellers, supply chain interviews, OEM management commentary, comparisons between earnings forecasts and market consensus, as well as historical P/E ratios and scenario valuations, to cross-validate demand durability, the extent of procurement pull-forward, deployment locations for AI workloads, and stock-level risk-reward.

Methodology notes

• Demand ResearchAlphaWise Enterprise Hardware Spending Survey

  Survey-driven demand trend assessment

  Assesses the growth rate and inflection points of spending on servers, storage, and personal computers through budgets, procurement timing, and capacity planning from hardware experts and CIOs.

• Industry ValidationChannel and Supply Chain Cross-Validation

  Multi-source fundamental verification

  Combines feedback from enterprise channels, value-added resellers, supply chains, and OEMs to validate changes in pricing, supply, market share, and procurement behavior.

• Valuation methodsP/E Ratio and Segment Valuation

  Relative valuation and sum-of-the-parts

  Compares next-fiscal-year P/E ratios with historical ranges and peers, and assigns valuation multiples to some companies separately for networking, AI servers, and traditional businesses.

• Risk-RewardBull, Base, and Bear Scenario Analysis

  Scenario valuation

  Evaluates target price ranges and risk-adjusted potential returns based on different combinations of earnings and valuation multiples.

Asset mapping & comparison

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

• Hewlett Packard Enterprise（HPE.N）

  Upgraded to Overweight, target price $69

  Strengths

  Servers, storage, and networking businesses all benefit from enterprise infrastructure spending; after the Juniper acquisition, networking business growth and margins improve, while cash flow and deleveraging prospects are favorable.

  Weaknesses

  A rising share of server revenue may dilute overall gross margin, and the target price is lowered from $71 to $69.

  Comparison

  Compared with DELL, server exposure purity is lower, but valuation support, networking business value, and re-rating potential are more attractive.

  Risks

  Acquisition integration, execution falling short of expectations, cyclical slowdown after demand pull-forward, and margin improvement failing to meet expectations.

• Everpure（P.N）

  Upgraded to Overweight, target price $108

  Strengths

  Has exposure to storage demand, market share gains, relatively attractive valuation, and a potential catalyst from a second hyperscale customer.

  Weaknesses

  Growth and re-rating depend to some extent on adding a new hyperscale customer and continued share gains.

  Comparison

  Compared with NTAP, risk-reward is more attractive, and the option value from hyperscale customer selection has not yet been fully priced by the market.

  Risks

  Delays in customer wins, storage momentum falling short of expectations, intensified competition, and failure of valuation re-rating.

• NetApp（NTAP.O）

  Upgraded from Underweight to In-Line with the market, target price $173

  Strengths

  Improving storage fundamentals support upward revisions to revenue and earnings forecasts.

  Weaknesses

  Valuation is already at a high P/E level, and many positives may already be reflected.

  Comparison

  Fundamental improvement is clear, but risk-reward is inferior to P.

  Risks

  Decline in storage demand, insufficient earnings upgrades, and compression of high valuations.

• Teradata（TDC.N）

  Downgraded from Overweight to In-Line with the market, target price $29

  Strengths

  AI-related opportunities remain attractive over the long term.

  Weaknesses

  The expected growth acceleration and earnings upgrades have not yet materialized, and there is insufficient basis for further valuation expansion.

  Comparison

  Compared with server and storage OEMs, near-term fundamental catalysts are weaker.

  Risks

  Continued delay in growth, earnings forecast downgrades, and contraction in valuation multiples.

• Dell Technologies（DELL.N）

  Maintained In-Line with the market, target price $430

  Strengths

  Leads in supply chain execution, with share and earnings upside potential in traditional servers, AI servers, and storage.

  Weaknesses

  The stock price has already been significantly re-rated, and there is uncertainty over the sustainability of procurement pull-forward, supply-driven share gains, and current profitability.

  Comparison

  It is the most direct beneficiary of server momentum, but valuation already reflects the recent strong fundamentals fairly fully.

  Risks

  AI server supply constraints, decline in server gross margin, peak earnings pullback, and valuation compression.

• STX and Wiwynn

  Maintained as preferred hardware names in the U.S. and Greater China, respectively

  Strengths

  STX benefits from hard disk drive and storage demand, while Wiwynn has high exposure to AI servers.

  Weaknesses

  Sensitive to hardware capital expenditure and supply-demand cycles.

  Comparison

  Compared with traditional personal computer vendors, they have higher exposure to growth in data centers and AI infrastructure.

  Risks

  Capital expenditure slowdown, reversal in supply-demand, pricing pressure, and cycle peaking.

• Lenovo

  Preferred name among traditional Asian OEMs

  Strengths

  Higher storage exposure, growing AI server project backlog, pricing power, and room for margin improvement in the infrastructure business.

  Weaknesses

  Still has exposure to the personal computer business, and the extent to which revenue forecasts lead consensus is limited.

  Comparison

  Compared with Acer, it has higher server and storage exposure and a clearer path to margin improvement.

  Risks

  Decline in personal computer demand, AI server progress falling short of expectations, and delayed margin improvement.

Key data

• Share of 2026 hardware budgets being pulled forward70% to 75%CIOs are pulling forward spending on personal computers, storage, and servers.
• Share of 2027 budgets being pulled forward20% to 30%Pull-forward purchasing is most pronounced in servers and personal computers.
• Share of respondents increasing on-premises compute capacityAbout 60%Mainly to meet AI workload demand.
• Share of respondents increasing all-flash storage capacityAbout 60%Reflects storage demand being jointly lifted by AI, refresh cycles, and data growth.
• Share of AI inference expected to be performed in the public cloudMore than 70%The public cloud is still the main deployment location for AI inference over the next three years.
• 2027 server and storage survey growth rate180 to 340 basis points above the average level of the past 8 yearsThe survey results reached a historical high.
• CY26 earnings forecasts relative to consensusMedian 8% higherCovers U.S. and Asian hardware OEMs.
• CY27 earnings forecasts relative to consensusMedian 11% higherThe report expects room for earnings upgrades has not yet been fully reflected.
• Hardware stock gains since the beginning of 2025More than 100%Shows the industry is already in the later stage of the cycle.
• Overall P/E ratio of U.S. hardware stocksAbout 25xClose to twice the previous peak valuation.

Impact & implications

In the short term, rising memory prices, supply concerns, and AI capacity buildout are expected to continue driving orders, price realization, and earnings forecast upgrades, benefiting servers, storage, networking equipment, and related channel partners. However, procurement pull-forward will draw on future demand, with personal computers likely to fall back first, servers then slowing in 2027 to 2028, while storage may extend into 2028 or later. Investors should tactically hold enterprise infrastructure names with reasonable valuations, room for margin improvement, and market share gains, and turn cautious when the breadth of earnings forecast upgrades peaks.

Risks

• This round of demand is mainly driven by budget pull-forward and refresh cycles, which may draw down hardware spending in 2027 and beyond.
• The industry is already in the later stage of the cycle, and a peak in the breadth of earnings forecast upgrades may become a signal to turn cautious.
• Valuations of U.S. hardware stocks and traditional enterprise hardware are significantly above historical levels, creating valuation compression risk.
• More than 70% of AI inference is still expected to take place in the public cloud, so the contribution of enterprise on-premises AI deployment to hardware demand may be lower than optimistic expectations.
• Continued increases in memory prices may ultimately compress OEM gross margins or weaken end-market budgets.
• Supply constraints in AI servers and key components may delay revenue recognition and increase execution risk.
• Personal computer demand may be the first to weaken from the second half of 2026 to 2027.

What to watch

• Whether the breadth and pace of enterprise hardware earnings forecast upgrades peak.
• Whether 2027 server and storage budgets can sustain growth above historical levels.
• The extent of the pullback in personal computer spending in the second half of 2026 and in 2027.
• The inflection point for server demand in 2027 to 2028, and whether storage demand can extend beyond 2028.
• Changes in DRAM and NAND prices, supply allocation, and enterprise procurement pull-forward behavior.
• Whether the share of enterprise on-premises AI compute and storage capacity rises, and whether the share of public cloud AI inference declines.
• HPE's networking business growth, margins, cash flow, and deleveraging progress after integrating Juniper.
• Whether P can secure a second hyperscale customer, and whether its enterprise storage market share continues to rise.
• DELL's AI server supply, traditional server pricing, and gross margin sustainability.