AI demand plus “chip inflation” is accelerating enterprise hardware procurement, with storage preferred over servers and servers preferred over personal computers
AI summary card
AI demand plus “chip inflation” is accelerating enterprise hardware procurement, with storage preferred over servers and servers preferred over personal computers
Morgan Stanley upgraded its IT Hardware industry view, arguing that “fear of missing out on procurement” caused by rising memory prices and AI expansion will extend the infrastructure upcycle, but elevated valuations and procurement pull-forward mean investors should remain selective and tactical.
- Enterprises view memory shortages as a structural issue and are therefore choosing to purchase servers, storage, and personal computers earlier rather than waiting for prices to decline.
- The AlphaWise survey shows that 70% to 75% of IT decision-makers are pulling forward 2026 hardware spending, while 20% to 30% are also pulling forward 2027 spending.
- About 60% of respondents are increasing on-premise compute and all-flash storage capacity due to AI demand, but more than 70% of AI inference is still expected to be performed in the public cloud over the next three years.
- The research preference ranking is storage, servers, and personal computers; storage demand is expected to last the longest, while personal computer spending has already begun to weaken.
- HPE and Everpure were upgraded to Overweight, NTAP was upgraded to In-Line, and TDC was downgraded to In-Line.
Report interpretation
Overview
The report revises the previous view that component inflation would suppress enterprise hardware demand. The latest surveys, channel checks, supply chain interviews, and vendor commentary indicate that enterprises are concerned about further increases in memory prices and insufficient supply, and are accelerating procurement; at the same time, AI is driving incremental compute and storage capacity. Together, these two forces support revenue growth, pricing power, and upward earnings revisions for servers, storage, and networking equipment. However, the current strong demand is still mainly driven by device refreshes, procurement pull-forward, and infrastructure modernization, rather than a broad, long-term enterprise AI supercycle.
Core views
Industry fundamentals are stronger than market consensus, but the cycle is already in a later stage. Morgan Stanley expects revenue for its covered portfolio of U.S. and Asian hardware OEMs to be about 5% above consensus in CY26 and CY27, with EPS 9% to 12% above consensus; its median CY26 EPS forecast is 8% above consensus and CY27 is 11% above consensus. From an investment perspective, investors should avoid companies that rely solely on personal computer demand or whose valuations have already been fully rerated, and should prioritize companies with higher exposure to storage, servers, and networking, room for margin improvement, and valuation support.
Analysis framework
The report uses the AlphaWise enterprise hardware spending survey as its core, combined with IT decision-maker surveys, enterprise channel checks, supply chain interviews, vendor management commentary, comparisons of earnings forecasts with market consensus, and historical P/E and risk-adjusted return analysis, to cross-validate demand durability, the degree of procurement pull-forward, end-market rankings, and individual stock valuations.
Methodology notes
Assess server, storage, and personal computer budgets, procurement timing, and AI capacity demand through enterprise hardware domain experts and IT decision-makers.
The survey is used to distinguish device refreshes, procurement pull-forward, and incremental AI demand, and to estimate the duration of growth and potential downturn years for different hardware categories.
Corroborate survey results with channel feedback, supply chain conversations, and OEM commentary.
This method is used to confirm whether memory price increases are accelerating procurement and to assess the competitive landscape and earnings leverage of servers, storage, and networking equipment.
Compare forward-year P/E ratios with historical ranges and use base, bull, and bear scenarios to measure risk-reward.
While raising earnings forecasts, the report lowers some valuation multiples to reflect elevated industry valuations, rising cycle maturity, and risks to earnings sustainability.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- HPEPreferred enterprise infrastructure OEM, upgraded to Overweight
- Strengths
- Exposure to servers, storage, and networking businesses; improved network business growth and margin potential after the Juniper acquisition; better prospects for free cash flow, deleveraging, and capital returns.
- Weaknesses
- A higher server revenue mix may dilute the overall gross margin.
- Comparison
- Compared with DELL, valuation expansion potential and network business value are more attractive, but execution capability still needs to be validated.
- Risks
- Integration and execution below expectations, server cycle peaking, and margin improvement failing to materialize.
- Everpure(P)Preferred storage name, upgraded to Overweight
- Strengths
- Benefits from improved storage spending, market share gains, attractive valuation relative to peers, and a potential catalyst from a second hyperscale customer.
- Weaknesses
- The investment thesis partly depends on adding a new hyperscale customer and continued share gains.
- Comparison
- Compared with NTAP, risk-reward is more attractive, and valuation reflects less of the potential growth catalyst.
- Risks
- Customer expansion below expectations, slowing storage demand, and reversal of the pricing and supply environment.
- NTAPStorage fundamentals improving, rating upgraded from Underweight to In-Line
- Strengths
- Storage demand has relatively strong durability, and earnings forecasts have room for upward revision.
- Weaknesses
- The P/E ratio is near a high level, and many positives have already been priced in.
- Comparison
- Fundamental improvement is similar to Everpure, but valuation risk is higher and risk-reward is relatively weaker.
- Risks
- Valuation pullback, insufficient earnings upgrades, and weakening enterprise storage budgets.
- DELLBeneficiary of servers and AI infrastructure, maintained at In-Line
- Strengths
- Leading supply chain execution, share gains in traditional servers and AI servers, and strong recent revenue and earnings growth.
- Weaknesses
- The stock price has already been significantly rerated, and there is uncertainty over whether procurement pull-forward and supply-driven share gains can be sustained.
- Comparison
- Server cycle earnings leverage is stronger than most peers, but valuation already reflects fundamental advantages relatively fully.
- Risks
- AI server supply bottlenecks, earnings peaking, valuation multiple compression, and share loss.
- STX and WiwynnTop picks in U.S. and Greater China hardware coverage, respectively
- Strengths
- STX benefits from the hard disk drive and storage cycle, while Wiwynn has high exposure to AI servers.
- Weaknesses
- Sensitive to hardware capital expenditure and cycle changes.
- Comparison
- Compared with traditional personal computer vendors, both have more direct exposure to growth in storage or AI infrastructure.
- Risks
- Slowing capital expenditure, reversal in component supply-demand, and earnings forecasts peaking.
- Personal computer-related namesLowest end-market preference
- Strengths
- May still receive near-term support from procurement pull-forward.
- Weaknesses
- Spending has already begun to deteriorate and is expected to enter a downturn earlier than servers and storage.
- Comparison
- Demand durability is weaker than storage and servers, and the report is cautious on HPQ, LOGI, and Acer.
- Risks
- Demand decline in the second half of 2026 to 2027, low-end market competition, and valuation compression.
Key data
- Procurement pull-forward ratio70% to 75%Surveyed IT decision-makers are pulling forward 2026 spending on personal computers, storage, and servers.
- 2027 spending pull-forward ratio20% to 30%Procurement pull-forward is most evident in servers and personal computers.
- AI on-premise capacity expansionAbout 60%About 60% of respondents are increasing on-premise compute capacity due to AI demand, with a similar proportion increasing all-flash storage capacity.
- Public cloud share of AI inferenceMore than 70%Most AI inference workloads are expected to continue running in the public cloud over the next three years.
- Server and storage growth expectations180 to 340 basis points above the average of the past eight years of surveys2027 growth expectations reached a historical high for the survey.
- Earnings forecast lead versus consensusCY26 median 8% above consensus, CY27 11% above consensusCovers U.S. and Asian hardware OEMs.
- HPE target price$69Rating upgraded from In-Line to Overweight; based on 15x FY27 EPS of $4.58.
- Everpure target price$108Rating upgraded from In-Line to Overweight, with the target price raised 24.1% from the previous $87.
- NTAP target price$173Rating upgraded from Underweight to In-Line, with the target price raised 26.3% from the previous $137.
- DELL target price$430Maintained at In-Line; earnings forecasts were raised, but the target price was lowered from $477, reflecting valuation multiple compression.
Impact & implications
In the near term, rising memory prices have not caused demand destruction; instead, they have increased enterprises’ willingness to lock in prices and supply, benefiting revenue, pricing, and earnings upgrades for servers, storage, networking equipment, and related channel partners. In the medium term, procurement pull-forward will draw down future budgets, with personal computers likely to weaken first in the second half of 2026 to 2027, servers potentially weakening in 2027 to 2028, and storage possibly continuing into 2028 or later. Therefore, the current environment is more suitable for holding high-quality companies with storage or server exposure, margin improvement, and valuation support, rather than indiscriminately chasing the entire hardware sector.
Risks
- Hardware stocks have risen by more than 100% cumulatively since the beginning of 2025, and the U.S. sector overall trades at about 25x P/E, with valuations significantly above historical levels.
- Procurement pull-forward may draw down 2027 and later budgets, causing revenue and earnings to decline rapidly in the later stage of the cycle.
- The scale of on-premise AI deployment may be lower than market expectations, because more than 70% of AI inference is still expected to be hosted by the public cloud.
- If memory prices or supply tightness ease earlier than expected, the demand acceleration brought by “fear of missing out on procurement” may reverse.
- The breadth of earnings forecast upgrades has already reached about 60%, and subsequent room for upward revisions may narrow.
- The cycle durability of servers and personal computers is weaker than storage, and differences in individual companies’ exposure to end markets may amplify drawdowns.
- Company execution, M&A integration, supply chain constraints, and margin improvement below expectations may weaken individual stock investment theses.
What to watch
- Whether earnings forecast upgrades for enterprise hardware have peaked, which is the core signal for shifting back to a cautious stance.
- Whether 2027 personal computer and server budgets flatten or decline as expected by the survey.
- Whether storage demand can continue into 2028 or later, and whether the storage cycle continues to lag servers by four to five quarters.
- Whether DRAM and NAND prices, supply allocation, and lead times continue to prompt enterprises to procure earlier.
- Changes in the actual distribution of AI inference workloads between on-premise and public cloud.
- HPE’s network business growth, margins, free cash flow, and deleveraging progress after integrating Juniper.
- Whether Everpure discloses a second hyperscale customer at its September 23 analyst day.
- Whether DELL’s AI server supply, shipment cadence, traditional server pricing, and share gains can be sustained.