Dell's $95 Billion AI Backlog Shows Agents Driving Traditional Server Demand

Dell Technologies reported on September 1 that it booked $60.9 billion in AI server orders in a single quarter and ended July with a $95 billion AI backlog. The headline numbers came from AI clouds and model builders, but the figure enterprise technology leaders should watch sits one line lower: Dell's traditional server and networking revenue more than doubled, and management tied that growth to AI workloads that need ordinary CPUs as much as GPUs. Three days later, HPE added its own record quarter. Together, the reports suggest the AI buildout is reaching the general-purpose data center, and that buyers face longer waits and higher prices across more of their hardware budget.

What did Dell report for Q2 fiscal 2027?

Revenue for the quarter ended July 31 was $47.0 billion, up 58% from a year earlier, according to Dell's earnings release. AI-optimized server revenue doubled to $16.4 billion. The Infrastructure Solutions Group, which sells servers, storage and networking, grew 89% to $31.8 billion, while the PC business grew 20% to $15.0 billion.

Dell raised its full-year revenue outlook by $25 billion, to about $192 billion. Vice Chairman and COO Jeff Clarke said in the release that IT has moved from a cost center to something that can "fuel growth and competitive advantage."

Adjusted earnings of $7.04 per share compared with a consensus of $4.92, and shares rose more than 6% in late trading, SiliconANGLE reported. The outlet also reported that Dell now expects AI server sales of about $74 billion for the year, and that Iren, an AI cloud provider, agreed during the quarter to buy $1.6 billion of Dell servers and networking gear.

Why are traditional servers suddenly growing?

Traditional servers and networking brought in $10.5 billion, up 122%, according to the release.

On the call, Clarke described "a growing trend of customers that require meaningful CPU compute capacity," SiliconANGLE reported. Holger Mueller of Constellation Research told the outlet that AI agents need CPUs as well as GPUs, which helps explain the demand. Agents spend much of their time calling tools, querying databases, running code and moving data, and most of that work runs on general-purpose processors, not accelerators.

AI Connect event speakers have made the same point from the buyer's side. In a AI Connect webinar on AI infrastructure at scale, a Rhino Federated Computing executive said his team splits preprocessing and postprocessing into CPU containers and reserves GPU machines for inference alone. Workload placement like that is becoming a basic cost discipline.

What does a $95 billion backlog mean for enterprise buyers?

The backlog is good news for Dell and a planning problem for everyone else. A queue that large means competition for the same components, and Dell is passing costs through. SiliconANGLE reported that Dell raised prices to offset rising component costs. Dell's balance sheet shows inventory roughly doubling since the start of the fiscal year, from $10.4 billion to $21.3 billion, as it builds against orders.

For enterprises, three implications follow:

• Hardware refreshes need longer lead times. If AI clouds and model builders are absorbing supply, general-purpose server orders placed late in the budget year may slip.
• Price assumptions from last year are stale. Capital plans for on-premises AI and for ordinary refreshes should reflect higher component costs.
• Idle capacity is expensive. Kaycee Lai, VP of AI at Everpure, warned at The AI Enterprise Conference 2026 that when storage can't keep up and GPUs sit idle, "that's not a good thing to tell your CFO." His argument for bringing AI to where the data already lives depends on the whole system, not just the accelerators.

Naveen Chhabra of Forrester told SiliconANGLE that the results suggest enterprise AI spending is moving past experimentation. Dell did not break out how much AI demand comes from ordinary enterprises versus AI clouds and model labs, so that share remains hard to measure.

How does HPE's quarter compare?

HPE reported fiscal third-quarter revenue of $12.2 billion on September 4, up 34% year over year, according to StorageNewsletter's summary of the release. Server revenue grew 35% to $6.8 billion and networking grew 75%, helped by the Juniper acquisition. HPE said its order backlog is at a record level but did not give a separate AI figure. CEO Antonio Neri said "AI is becoming a multi-year growth driver for HPE."

HPE's growth is slower than Dell's, but the pattern matches: AI demand is spilling into servers and networks that enterprises already buy.

What to watch next

The question for the rest of 2026 is whether enterprise buyers, not just AI clouds, convert pilots into hardware orders, and whether supply loosens enough for them to get it on time. Leaders planning private AI should match model size to the job, as Mastercard's Pushkar Rege advised in a panel on AI cost per outcome, and budget for CPUs, storage and networking alongside GPUs. The agent era looks less like a GPU purchase and more like a full data center refresh.

Frequently asked questions

How big is Dell's AI server backlog?

Dell reported a $95 billion AI server backlog at the end of its fiscal second quarter, ended July 31, 2026, after booking $60.9 billion in AI orders in the quarter.

Why are traditional server sales growing so fast?

Dell executives said more customers need significant CPU capacity for AI, and analysts noted that AI agents rely on CPUs for tool calls, data movement and code execution as well as GPUs.

What do Dell's results mean for enterprise IT budgets?

Enterprises should plan for longer hardware lead times and higher component-driven prices, and budget for CPUs, storage and networking alongside GPUs.

Filed under: Data & AI Infrastructure, AI Strategy & ROI