Feedback from a large U.S. VAR shows enterprise demand for hardware, networking, and AI infrastructure is stronger than planned
AI summary card
Feedback from a large U.S. VAR shows enterprise demand for hardware, networking, and AI infrastructure is stronger than planned
J.P. Morgan channel interviews indicate broad-based strength in enterprise spending across networking, data centers, cybersecurity, and AI, with order backlogs extending into 2H26 and 2027, while rising memory prices and longer delivery cycles are increasing OEM pricing and supply risks.
- The interviewed VAR said demand is above plan, with revenue growth significantly exceeding the usual double-digit annual growth target, with strength across networking, data centers, cybersecurity, and AI.
- The 12-month order backlog has extended into 2H26 and into 2027 and was described as unprecedented; some customers are placing larger orders earlier to lock in supply, but the VAR believes this reflects real demand.
- Rapidly rising memory prices are affecting storage and compute the most; some OEM quote validity periods have shortened from the previous 60-90 days to 1-2 weeks, and some even retain the right to reprice at shipment.
- Server and storage demand is broadly strong, with Dell seen as standing out in AI factories; AI-oriented storage vendors such as Vast and Weka are also gaining traction.
- In enterprise edge networking, Cisco is seeing strong adoption due to an improved product portfolio, while HPE/Juniper are the main alternatives; in AI networking, Nvidia is very strong, and white-box solutions are gaining share among hyperscale customers.
- Localization repatriation and private data center investment are accelerating, driven by cloud billing pressure, concerns about putting proprietary or sensitive data in the cloud, and changing cost comparisons for workloads with large-scale token consumption.
- F5 is seen as having at least the potential to match its record growth in 2025 and to benefit from API-driven application security and inter-agent traffic driven by MCP.
- Frontier model risks are bringing incremental budgets for AI, security, and infrastructure refreshes rather than clearly crowding out other budgets; at the same time, enterprises are beginning to govern token consumption in a manner similar to credit limits.
Report interpretation
Overview
This report is based on J.P. Morgan’s recent discussions with a large U.S. VAR and distills trends in enterprise IT demand, customer spending, OEM pricing and supply, AI infrastructure, networking, security, and enterprise software adoption. The core conclusion is that demand for enterprise hardware and networking remains in a strong expansion phase, particularly driven by AI, cybersecurity, data center buildouts, and repatriation to private infrastructure; at the same time, soaring memory costs and longer delivery cycles are becoming the main constraints at the supply chain and pricing levels.
Core views
The report argues that current enterprise demand is not simply driven by stockpiling or short-term early ordering, but is supported by real demand for AI, data centers, cybersecurity, and infrastructure refreshes. Servers, storage, networking equipment, and security infrastructure are all performing strongly, with Dell, Cisco, F5, Nvidia, and some AI storage and white-box vendors specifically cited as beneficiaries in different sub-segments. On the other hand, rising memory prices are pushing up compute and storage prices, with some OEMs shortening quote validity periods and potentially repricing at shipment, and supply constraints are expected to continue into 2H26 and 2027.
Analysis framework
The analytical method is channel interviews combined with mapping to covered companies: by using feedback from a large U.S. VAR on customer demand, order backlogs, supply cycles, OEM pricing, competitive vendor dynamics, AI applications, and budget governance, the report infers operating trends and potential investment implications for companies related to hardware, networking, security, cloud, and software.
Methodology notes
Observe enterprise IT procurement behavior through large value-added resellers
VARs sit between enterprise customers and OEM vendors and can observe order cadence, customer budgets, supply constraints, price changes, and vendor share shifts, making them suitable for validating demand trends in hardware, networking, and software.
Use backlog and lead time to judge demand strength and supply bottlenecks
The report uses signals such as the 12-month backlog extending into 2H26 and 2027 and storage and compute lead times exceeding six months as evidence of simultaneously strong demand and supply constraints.
AI adoption drives spending on servers, storage, networking, security, observability, and token governance
The report focuses not only on AI compute equipment, but also on adjacent budget expansion driven by frontier model risks, agentic AI, API security, private data centers, and AI FinOps.
Asset mapping & comparison
Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).
- Dell Technologies (DELL)Beneficiary of AI servers and AI factories
- Strengths
- The VAR said Dell stands out among major OEMs, especially in AI factories.
- Weaknesses
- Storage and compute are most exposed to rising memory prices and delivery cycles exceeding six months.
- Comparison
- Compared with HPE, Cisco, SuperMicro, and Lenovo, Dell was specifically described as a standout vendor in AI servers.
- Risks
- Memory costs, repricing mechanisms, and supply lead times may affect order fulfillment, customer budgets, and margins.
- Hewlett Packard Enterprise (HPE) / JuniperAlternative option for server adoption and enterprise edge networking
- Strengths
- HPE was mentioned as seeing adoption on the server side, and HPE/Juniper are the main alternatives to Cisco in enterprise edge networking.
- Weaknesses
- The VAR has not yet seen a meaningful revenue inflection from the HPE/Juniper data center portfolio.
- Comparison
- Cisco is stronger in the enterprise edge due to product portfolio improvements; HPE/Juniper are technically viable but commercial momentum has not yet clearly accelerated.
- Risks
- Uncertainty around integration execution, customer willingness to switch, and data center revenue conversion.
- Cisco (CSCO)Beneficiary in enterprise edge networking, data center networking, and Nvidia partnership
- Strengths
- The enterprise edge is seeing strong adoption due to the refreshed product portfolio; Cisco’s Nvidia partnership is also attracting customer interest.
- Weaknesses
- There are share shifts with Arista in traditional data center networking, while Nvidia and white-box solutions are performing strongly in AI networking.
- Comparison
- Cisco is stronger in the enterprise edge; in data center networking it competes with Arista, and in AI networking it faces Nvidia and white-box solutions.
- Risks
- Competition for AI networking share, the white-box trend, and customer architecture changes may affect growth.
- Arista (ANET)Beneficiary of traditional data center networking
- Strengths
- The VAR said Arista and Cisco are both seeing share changes in traditional data center networking.
- Weaknesses
- The report did not indicate equally broad positive feedback for Arista in enterprise edge or AI networking.
- Comparison
- It competes directly with Cisco for traditional data center networking share; in AI networking, Nvidia and white-box solutions were described as stronger.
- Risks
- Share volatility, rising white-box share among hyperscale customers, and changes in AI networking architecture.
- Nvidia (NVDA)Beneficiary of AI networking and AI cluster demand
- Strengths
- The VAR said Nvidia is very strong in AI networking, and Cisco’s Nvidia partnership is also attracting customer interest.
- Weaknesses
- AI cluster lead times have already exceeded six months, and supply constraints may limit near-term deliveries.
- Comparison
- In AI networking it holds a strong position relative to traditional networking vendors and white-box solutions.
- Risks
- Supply-demand mismatches, longer lead times, and customer budget governance may affect near-term revenue cadence.
- F5 Inc (FFIV)Beneficiary of API security and agentic AI traffic security
- Strengths
- The VAR said F5 is at least likely to match its record growth in 2025; its positioning has gained recognition as it shifts from load balancing to protecting API-driven applications.
- Weaknesses
- Growth depends on continued realization of trends in API security and inter-agent traffic driven by MCP.
- Comparison
- The report does not provide a direct quantified peer comparison, but positions F5 as a beneficiary of agentic AI connection security.
- Risks
- Evolution in AI application architecture, customer security budget allocation, and competing solutions may affect growth durability.
- Microsoft (MSFT)Default option for enterprise AI applications
- Strengths
- The VAR believes M365 Copilot is the default customer choice because of the installed base of 365, and it is used for enterprise data access, email, calendars, and workflow automation.
- Weaknesses
- Customers are also seriously evaluating alternatives such as Claude Cowork and Gemini Enterprise.
- Comparison
- There is currently no sign that alternative solutions are driving customers to replace Microsoft Copilot usage.
- Risks
- Tighter token governance, maturation of competing suites, and changes in licensing bundle strategy may affect the growth slope.
- Anthropic / Claude CoworkEnterprise AI alternative suite and one source of frontier model risk
- Strengths
- Claude Cowork was described positively by the VAR; Mythos-class frontier models have become a focus of customer attention.
- Weaknesses
- The report says there is still no sign that alternative solutions are replacing Microsoft Copilot usage.
- Comparison
- Compared with Microsoft Copilot, Claude Cowork is a competing suite being seriously evaluated but has not yet clearly displaced the default option.
- Risks
- There is still uncertainty around enterprise adoption, governance, security risks, and budget approvals.
- Alphabet / Gemini EnterpriseCompeting enterprise AI suite
- Strengths
- Customers are seriously evaluating alternatives such as Gemini Enterprise.
- Weaknesses
- The report does not show it replacing Microsoft Copilot.
- Comparison
- Like Claude Cowork, it is an enterprise AI alternative outside Microsoft Copilot.
- Risks
- Enterprise switching costs, stickiness of the existing Microsoft 365 ecosystem, and token budget controls.
- NetApp (NTAP), Vast, Weka, EverpureBeneficiaries of AI-oriented and traditional storage demand
- Strengths
- AI-oriented storage vendors are gaining traction, while Dell, Everpure, and NetApp remain strong; traditional storage vendors had already surpassed last year's full-year performance by July.
- Weaknesses
- Storage is the most exposed to rising memory costs and longer delivery cycles.
- Comparison
- AI-oriented vendors such as Vast and Weka are showing clear momentum, while traditional vendors remain strong.
- Risks
- Memory prices, supply constraints, customer budgets, and changes in AI storage architecture choices.
Key data
- Demand performanceAbove planThe VAR said revenue growth is significantly above the usual double-digit annual growth target, with strengths across networking, data centers, cybersecurity, and AI.
- Order backlogExtended to 2H26 and 2027The 12-month order backlog was described as unprecedented, with some customers locking in supply through larger orders.
- OEM quote validity period1-2 weeksSome OEM quote validity periods have shortened from the previous 60-90 days to 1-2 weeks, and in some cases they retain the right to reprice at shipment.
- Lead timeMore than six monthsLead times for storage, compute, and AI clusters exceed six months; networking is relatively less affected because it uses less memory.
- Duration of supply constraintsExpected to continue into 2H26 and 2027The VAR expects memory-related supply constraints to persist in 2H26 and 2027.
- Private data center spendingAlready above last year's level in 2026Cloud costs, concerns over sensitive data, and the economics of large-scale token consumption are driving repatriation to on-premise environments.
- J.P. Morgan Global Equity Research coverage rating distributionOverweight 53%；Neutral 36%；Underweight 12%As of July 4, 2026, the table is disclosed for rating distribution explanation and does not represent a new rating on any single company in this report.
Impact & implications
In terms of investment implications, the report supports a positive view on the resilience of demand for enterprise hardware, networking, security, and AI infrastructure, especially benefiting vendors exposed to AI factories, data center networking, private cloud/on-prem repatriation, security, and observability budgets. On pricing, rising memory costs may support OEM revenue and pricing, but they also bring customer budget pressure, order timing volatility, and gross margin uncertainty. On the software side, Microsoft Copilot remains the default enterprise option, but alternative suites such as Claude Cowork and Gemini Enterprise are being seriously evaluated; token governance is shifting from loose control toward a credit-limit model, which may suppress the slope of consumption revenue in the short term but helps the sustainability of spending.
Risks
- Rapidly rising memory prices may continue to push up server and storage costs and force OEMs to shorten quote validity periods or reprice at shipment.
- Lead times for storage, compute, and AI clusters exceed six months, which may delay revenue recognition, customer deployments, and order fulfillment timing.
- Although customers ordering early to lock in supply is viewed by the VAR as real demand, it may still bring forward order timing and create future comparison-base pressure.
- Enterprise token governance is shifting from loose control toward a credit-limit model, which may constrain short-term consumption revenue.
- Share shifts among Nvidia, white-box solutions, Cisco, Arista, and HPE/Juniper in AI networking may intensify competitive uncertainty.
- This report is based on interviews with a single large U.S. VAR; the conclusions represent important channel feedback but are not equivalent to complete market-wide statistics.
What to watch
- Whether lead times for servers, storage, and AI clusters continue to lengthen in 2H26 and 2027.
- Changes in memory’s share of OEM total cost, and whether OEM quote validity periods and shipment repricing terms continue to tighten.
- After enterprise customers place orders early, whether actual deployment and consumption match the order backlog.
- Whether private data center repatriation continues to be driven by cloud bills, sensitive data concerns, and the economics of token costs.
- Whether Cisco’s product cycle in enterprise edge networking can continue, and whether the HPE/Juniper data center portfolio shows a revenue inflection.
- Whether F5’s growth in API security, MCP, and agentic AI connection security scenarios continues.
- Whether Microsoft Copilot maintains its default-on status, and whether alternatives such as Claude Cowork and Gemini Enterprise begin to affect actual usage share.
- The short-term suppression and long-term sustainability impact of AI FinOps and token credit-limit governance on enterprise AI consumption revenue.