Dell Projects AI Will Account for 75 Percent of Datacenter Demand by 2030

Dell’s AI Business and Server Refresh Cycle Accelerate

Dell’s AI business is expanding as GPU accelerator supply improves, while a long-delayed server replacement cycle is also gaining momentum. Companies are modernizing application and database server fleets to free power and datacenter space for energy-intensive AI systems.

During a conference call covering Dell’s second quarter of fiscal 2027, Jeff Clarke, the company’s chief operating officer and vice chairman, said Dell took eight quarters to reach 3,200 customers that had purchased an “AI factory.” The next 3,300 customers were acquired in only three quarters, bringing the total to 6,500 organizations that have deployed AI systems based on Dell PowerEdge products.

Many of these customers are Tier 2 cloud providers, service providers, neoclouds, and sovereign organizations. Large enterprises are also turning to Dell, their traditional server supplier, for both infrastructure upgrades and AI clusters used for inference and potentially training.

Dell reported more than 1.2 million installed servers from the PowerEdge 14G generation or earlier. The 14G systems were announced in 2017 and entered production in 2018, making much of this equipment roughly eight years old or older. Dell’s current 17G systems and forthcoming 18G systems can consolidate these older machines substantially.

Clarke estimated that six to eight older servers can be consolidated into one 17G system, while 12 to 14 can be replaced by one 18G system. These newer systems are often more heavily configured and therefore more expensive. Rising costs for DRAM, flash, CPUs, and other components are also contributing to higher traditional-server revenue.

Strong Quarterly Results

For the quarter ended July 31, Dell recorded $41.11 billion in product sales, an increase of 71.8 percent from the same period a year earlier and 7.9 percent sequentially. Services revenue was flat at $5.86 billion.

Combined product and services revenue reached $46.97 billion, up 57.7 percent year over year. Dell attributed the results in part to maximizing CPU, GPU, DRAM, and flash allocations, improving its large-scale manufacturing operations, and increasing PC margins.

Operating income rose 3.1 times to $5.36 billion. Net income reached $4.13 billion, up 3.55 times year over year and equal to 8.8 percent of revenue. This was Dell’s highest profitability level in the coverage period since the Great Recession, excluding Q2 F2020 and Q3 F2022, when company divestitures contributed significantly to results.

Dell’s debt load has started to rise again, although it remains well below the levels reached after the company acquired EMC and VMware a decade ago and later went private. Debt fell below $30 billion after VMware was spun out and had been declining for several years. Cash and debt are now increasing at approximately similar rates, leaving relative net debt largely unchanged despite the investment required for AI systems.

Michael Dell and his wife own 41 percent of Dell’s shares and receive the same proportion of dividend payments. Dell distributed $2.2 billion in cash during the second fiscal quarter and held $14.25 billion in cash and investments. Revenue growth continues to outpace debt growth.

Server, Networking, and Storage Results

Dell’s server and networking business, which consists primarily of servers, generated $26.93 billion in revenue, up 2.1 times year over year and 9.1 percent sequentially. Storage revenue, built on Dell’s EMC foundation, rose 25.8 percent to $4.85 billion.

Dell also provided an update on Project Lightning, a parallel file system designed for high-end AI workloads. The system is intended to compete with Lustre, controlled by DataDirect Networks, as well as storage products from VAST Data, WEKA, and Everpure, formerly Pure Storage.

Lightning is separate from Isilon’s OneFS, the high-performance computing file system used in Dell PowerScale systems for managing very large files. Lightning is designed instead for workloads involving large numbers of small input and output operations against flash-based block storage.

“Lightning, our parallel file system designed for native AI use cases, we continue to have the product out in the field,” Clarke said. “We continue to see interest. It is still a relatively new product. It is in beta at several customers. We are in runoffs against other competitors with the product. That will continue. And as it builds momentum, I am certain we will give you an update in the future. But that's where Lightning is at the moment.”

Dell’s Infrastructure Solutions Group reported $31.78 billion in sales, up 89.2 percent year over year, with operating income of $4.78 billion, up 3.25 times. Dell’s PC business generated approximately $15 billion in revenue, up 20.2 percent, and $1.14 billion in operating income, up 42.2 percent.

AI Systems Surpass Traditional Servers

Dell has reported AI systems separately from its traditional server and networking business for the past two quarters. Its AI systems business was larger than its traditional server business a year ago, and the difference remained substantial. However, traditional server sales accelerated during the quarter and could exceed AI server sales in Q4 F2027.

Dell sold $16.4 billion in AI systems during the quarter, twice the amount from a year earlier and nearly equal to the Q1 F2027 figure. Traditional servers and networking generated $10.53 billion, an all-time quarterly high and 2.2 times the year-ago result.

Dell ended the quarter with $95 billion in AI-system bookings, including $60.9 billion booked during the quarter. The backlog was 8.1 times larger than it was a year earlier.

Dell expects approximately $19 billion in AI-system sales in Q3 and another $22.5 billion in Q4, which would bring full-year AI-system revenue to $74 billion. Based on Infrastructure Solutions Group forecasts and estimates for traditional-server revenue, AI-system sales are projected at $15.6 billion in Q3 and $25.1 billion in Q4. Traditional servers are therefore expected to surpass AI servers in Q4, as Nvidia’s “Vera-Rubin” systems ramp toward full-volume production.

Dell’s 2030 AI Demand Forecast

Clarke said Dell expects agentic AI workloads to reshape datacenter infrastructure. He stated that inference has surpassed training as a source of demand and projected that inference-driven token volume will grow 87 times to 3,600 quadrillion tokens by 2030. Training demand is expected to increase fivefold to 850 zettaflops during the same period.

Enterprise agentic AI is expected to become the largest workload by 2028. Dell forecasts that AI will represent 75 percent of total datacenter demand by 2030 and add 200 gigawatts of power demand over that period. Clarke said half of this opportunity is expected to come from neoclouds, sovereign organizations, and enterprises.

Dell estimates the overall opportunity at more than $1 trillion through 2030. The company is competing with original design manufacturers and other equipment vendors, including Supermicro, IEIT Systems, Sanmina, Quanta Computer, and Foxconn. Sanmina acquired the ZT Systems business from AMD.

Dell’s position in the AI infrastructure market reflects both its manufacturing capacity and its access to Nvidia GPU allocations. Those systems are being supplied to neoclouds, sovereign organizations, and large enterprises, while hyperscalers and cloud builders increasingly develop their own XPU accelerators alongside purchasing GPUs from Nvidia.