Barclays Raises Server Model: Enterprise Beats Expectations, AI Cloud Demand Remains the Main Theme

Overview

Following recent earnings and industry discussions, Barclays updated its server model, shifting from the previous AI vs. Traditional framework to Cloud (including AI) and Enterprise servers. The core conclusion is that Cloud servers continue to be driven by upward revisions in hyperscaler capital spending, while Enterprise servers materially outperformed expectations this quarter, mainly due to higher ASPs, configuration upgrades, and some unit volume growth.

Core views

The report remains constructive on the long-term momentum of AI servers and Cloud servers, noting that hyperscaler CY26 capital spending expectations have risen to more than $700.0 billion, well above about $423.0 billion in CY25. Near-term Enterprise server performance is also strong, with the CY26 revenue growth forecast revised up to about 33%, but Barclays believes this growth is unlikely to be sustained because further price increases may suppress demand, and this year's pull-forwards and ASP contribution will create a higher base; therefore, it assumes only about 1% growth in CY27 with downside bias.

Analysis framework

The report uses a top-down capital spending and server market model, combined with the latest results, orders, revenue guidance, and management commentary from OEMs such as DELL, HPE, and SMCI, to update Cloud and Enterprise server revenue, growth, ASP, unit volume, and share changes.

Methodology notes

• Market segmentationCloud and Enterprise server split

  Cloud servers are defined as cloud demand including AI servers, while Enterprise servers are defined as enterprise-side traditional and inference-related server demand.

  Barclays adjusted its modeling approach this year, no longer using the old AI vs. Traditional split, and instead estimating Cloud and Enterprise separately to better reflect demand differences across hyperscaler cloud capex, enterprise inference, agentic AI, and traditional servers.

• Growth driver decompositionASP and unit volume contribution

  Breaks revenue revisions into ASP increases, pull-forwards, configuration upgrades, and unit volume growth.

  The report believes most of the upward revision for Enterprise servers comes from ASPs and pull-forwards, but companies such as DELL also show some growth in traditional server unit volumes this quarter; if prices continue to rise, demand elasticity may weaken in the future.

• Scenario riskPace of AI build-out and revenue recognition visibility

  AI training, sovereign AI, and inference projects may provide upside, but revenue may also be delayed due to funding, deployment, or customer readiness.

  Barclays notes that its CY27 cloud server forecast may still have upside bias, but financing for large AI training projects, the deployment pace of Middle East sovereign AI, and SMCI's revenue recognition timing could create quarterly volatility.

Asset mapping & comparison

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

• DELL

  A key beneficiary of AI servers and traditional Enterprise servers

  Strengths

  AI server revenue accounted for about 37% of total quarterly revenue, up 80% quarter-over-quarter; AI backlog reached $51.3 billion; FY AI server revenue guidance was raised to about $60.0 billion; traditional servers grew more than 90% year-over-year.

  Weaknesses

  Part of traditional server demand may come from price increases and pull-forwards, so sustainability still needs verification; rising commodity costs may create pressure.

  Comparison

  Among branded OEMs, DELL is identified in the report as an important contributor to upward revisions in Enterprise and AI servers, with full-year traditional server growth expected to exceed 60%.

  Risks

  Slower customer AI investment, intensifying AI competition, delayed PC recovery, slowing non-AI traditional business, and rising commodity costs.

• HPE

  A beneficiary related to Enterprise servers, AI orders, and networking business

  Strengths

  Traditional server orders grew triple digits year-over-year, with quarterly AI orders of $1.8 billion and cumulative orders of $16.4 billion; management sees agentic AI and inference use cases driving enterprise project launches.

  Weaknesses

  Current growth is heavily supported by ASP, and AI deal selection is more cautious with a focus on gross margin and working capital.

  Comparison

  After acquiring JNPR, Barclays is more inclined to view HPE as a networking company rather than a pure server company, because networking is expected to contribute more than 50% of full-year corporate operating profit.

  Risks

  Sustainability of ASP-driven demand, pace of converting AI orders into revenue, and pressure on gross margin and working capital.

• SMCI

  An important participant in the AI server and white-box server chain

  Strengths

  Enterprise revenue share rose from about 15% in the previous quarter to about 28% this quarter, and contributed better-than-expected margins; the company still has record backlog.

  Weaknesses

  Last quarter revenue of $10.2 billion was below guidance of at least $12.3 billion, with customer readiness affecting revenue; visibility into revenue recognition is limited.

  Comparison

  Relative to DELL and HPE, SMCI is more exposed to AI project delivery timing, customer readiness, and volatility in white-box servers.

  Risks

  Customers slowing plans due to lawsuits and investigations, delayed revenue recognition, and volatility in the pace of AI build-out.

• NVDA-related white-box servers

  A key chain benefiting from share expansion in Cloud AI servers

  Strengths

  The report notes that white-box servers, including NVDA-related revenue, continue to gain market share from branded vendors, and that the Cloud server market is already significantly larger than the Enterprise server market by revenue.

  Weaknesses

  Once Enterprise inference and sovereign AI scale more visibly, some share may shift back toward branded vendors.

  Comparison

  In the short term, Cloud AI training demand is more favorable for white-box vendors and the NVDA ecosystem, while Enterprise and inference demand may be more favorable for branded OEMs.

  Risks

  Insufficient financing for large AI training deals, delayed deployment of sovereign AI projects, and volatility in the pace of AI server build-outs.

• Enterprise server market

  Driven jointly by rising ASPs, configuration upgrades, and enterprise inference demand

  Strengths

  CY26 revenue growth expectations were sharply revised up from about 2% to about 33%, and both DELL and HPE provided strong commentary on traditional servers.

  Weaknesses

  Growth is mainly driven by ASPs and pull-forwards, with only about 1% growth expected in CY27 and downside bias.

  Comparison

  Compared with Cloud servers, Enterprise server growth is more affected by pricing, procurement cycles, and enterprise demand elasticity.

  Risks

  Continued price increases eroding demand, unit volumes falling short of expectations, and this year's high base causing growth normalization next year.

Key data

• Hyperscaler capital spendingMore than $700.0 billion in CY26, about $423.0 billion in CY25This is a key indicator for AI server demand and Cloud server forecasts.
• Cloud server revenue forecastCY26 $567.0 billion; CY27 $885.0 billionThe previous forecasts were $585.0 billion and $817.0 billion, respectively; the new growth rates are 77% and 56%.
• Cloud server five-year CAGRAbout 62% for 2022-2027EHigher than the previous estimate of about 59%.
• Enterprise server revenue growthAbout 33% in CY26; about 1% in CY27The previous CY26 forecast was about 2%; for CY27, only low growth is assumed due to a high base and the risk of demand erosion.
• Total server revenue growth70%/49% in CY26/CY27The previous forecast was 71%/36%, reflecting an upward revision for CY27.
• Enterprise server ASP and unit volumeASP may be up nearly 40%-50%, while unit volume may grow by mid-single digits to high-single digitsThe report believes most of the revenue revision comes from ASP rather than pure unit expansion.
• DELL AI serversAI server revenue accounted for about 37% of total quarterly revenue, up 80% quarter-over-quarter; AI backlog was $51.3 billion, up 19% quarter-over-quarterManagement raised FY AI server revenue guidance from $50.0 billion to about $60.0 billion.
• HPE AI and traditional server ordersQuarterly AI orders of $1.8 billion, cumulative AI orders of $16.4 billion; traditional server orders up triple digits year-over-yearThe report believes HPE places greater emphasis on gross margin and working capital in AI deals.
• SMCI revenue and guidanceLast quarter revenue was $10.2 billion, below guidance of at least $12.3 billion; FQ4 revenue guidance is $11.0 billion-$12.5 billionFY26 revenue guidance is $38.9 billion-$40.4 billion, slightly below the previous statement of at least $40.0 billion.

Impact & implications

From an investment perspective, the AI infrastructure chain still has strong revenue momentum, with Cloud servers and white-box/NVDA-related servers continuing to benefit; at the same time, the short-term improvement in Enterprise servers supports valuations and earnings for branded OEMs such as DELL and HPE. However, if rising ASPs weaken demand, or if AI deployment and revenue recognition are delayed, CY27 growth expectations may face downward revisions.

Risks

• Limited visibility into AI server build-outs could lead to quarterly revenue volatility or delays.
• Large AI training projects require sufficient funding support; otherwise, Cloud server forecasts may fall short.
• If Middle East sovereign AI projects are delayed in deployment, revenue recognition may be pushed out.
• Enterprise server growth is mainly driven by ASPs and pull-forwards, and continued price increases could erode demand and unit volumes.
• Rising memory and commodity costs may compress gross margins and push prices higher.
• SMCI faces specific risks related to customer readiness, revenue recognition visibility, and slowdowns among investigation-related customers.
• Barclays discloses that it or its affiliates may have investment banking or other commercial relationships with covered companies, and investors should consider potential conflicts of interest.

What to watch

• Whether hyperscaler CY26 and CY27 capital spending guidance continues to be revised upward.
• The financing, deployment, and revenue recognition pace of AI training, inference, and sovereign AI projects.
• Enterprise server ASP increases, unit volume changes, and demand elasticity.
• Changes in AI backlog, order conversion to revenue, and management guidance at DELL, HPE, and SMCI.
• Share shifts between white-box/NVDA-related servers and branded OEMs.
• The impact of memory, commodity costs, and higher configurations on gross margins and customer budgets.