# AI Server Market Update: Vendors Shift from Silicon to Services — retained record

> Retained news record from news:datacenterknowledge.

Canonical: https://highsignal.app/data/records/7bfb44f58d96fdb2

[← sources](https://highsignal.app/data)


source: [news](https://highsignal.app/data/news) (news:datacenterknowledge)  published 2026-05-13 09:00 ingested 2026-05-14 08:36

[original source ↗](https://www.datacenterknowledge.com/servers/ai-server-market-update-vendors-shift-from-silicon-to-services)

## extract

AI Server Market Update: Vendors Shift from Silicon to Services
Dell, HPE, Lenovo, and Supermicro are riding record AI server demand, but winning enterprise customers requires more than just Nvidia chips.
The AI server market continues its explosive growth, fueled primarily by demand for GPUs – particularly from Nvidia. As the customer base broadens beyond hyperscalers and neoclouds to include enterprise buyers, hardware manufacturers face a new challenge: differentiation.
Unlike hyperscalers, which know precisely what they need and demand rapid, large-scale deployments, many enterprises are still determining how AI improves their business processes. They often lack the engineering skills to deploy AI infrastructure or manage the power and cooling demands of modern AI server racks. For server vendors, winning this enterprise business requires more than silicon. According to industry analysts and executives, success depends on offering comprehensive software management tools and professional services that guide organizations through every stage of their AI journey.
“You can’t just throw hardware at customers in the AI space. They need all the support and services they can get,” said Peter Rutten, an IDC research vice president, in an interview with Data Center Knowledge.
AI Infrastructure Market Growth
IDC reports the global server market reached a record $444 billion in 2025. With AI infrastructure remaining a strategic priority, IDC projects AI infrastructure spending will reach $487 billion in 2026 and surpass $1 trillion by 2029. Analysts add that demand continues to outstrip supply, with GPU availability still constrained and memory shortages driving up costs.
Recent earnings reports underscore the acceleration:
In fiscal Q4 FY2027 (ended Jan. 30, 2026; reported Feb. 27), Dell Technologies posted $9 billion in AI-optimized server revenue, up 342% year over year, and disclosed a $43 billion AI backlog.
Lenovo’s Q3 FY 2025/26 results (reported Feb. 12) cited high double-digit AI server revenue growth and an AI server pipeline of about $15.5 billion.
In Q1 FY2026 (reported March 9), Hewlett Packard Enterprise reported $4.2 billion in server revenue within its Cloud & AI segment and a $5 billion AI systems backlog, primarily from enterprise and sovereign AI customers.
Supermicro’s fiscal Q3 FY2026 (ended March 31; reported May 6) delivered $10.2 billion in sales, with more than 80% from AI GPU-related platforms; orders and backlog remain strong.
Why Differentiation Matters
While hyperscalers and neoclouds continue to drive AI infrastructure spending, analysts and executives say sovereign AI initiatives (national and regional programs to maintain data control and compliance) and enterprise customers represent the next growth wave. With vendors reporting massive backlogs and demand outpacing supply, some buyers prioritize access to Nvidia GPUs through their preferred OEM – including Dell, HPE, Lenovo, and Supermicro – over the nuances of server brand differentiation.
“Are people buying Dell servers? Are people buying Lenovo servers? Or are they really buying Nvidia servers that are being fulfilled through their OEM of choice?” said Matt Kimball, vice president and principal analyst at Moor Insights & Strategy, in an interview with Data Center Knowledge. “There’s an argument to be made that really what people want are more Nvidia chips. The servers are just really kind of bending metal around those Nvidia chips.”
Rutten echoed that view. “If you look at it, 90% of what they're selling they have in common,” he said. “They’re all selling systems that are capable of running Nvidia GPUs.”
At a large scale for neoclouds and sovereign AI initiatives, hardware differentiation is constrained by standardized reference designs and ecosystems. Nvidia’s designs and rack-level guidance shape much of the architecture, shifting vendor differentiation toward integration, delivery speed, power and cooling options, software tooling, and services. Vendors also offer AMD Instinct alternatives for specific buyers and workloads, but most large deployments today remain standardized on Nvidia platforms.
Many neoclouds are younger, smaller companies that need help installing the AI infrastructure, said Scott Patti, vice president of enterprise solutions for the Lenovo Infrastructure Solutions Group. “There’s a lot of joint engineering collaboration in all those deployments because they are of very significant scale,” Patti told Data Center Knowledge.
In contrast, hyperscalers generally do not need that assistance; they use their own engineering teams and custom management tools, Kimball noted.
The Enterprise Opportunity
For server vendors, the enterprise segment is where differentiation matters most. Enterprises aren’t looking for raw GPU horsepower alone – they want solutions to specific business problems, Patti said.
“Anyone can bring them a box with a GPU or CPU,” he said. “AI is not one-size-fits-all. Our services and deployment teams are critical to build that relationship and understand what customers are focused on – content creation, customer support automation, code development – and then we can match that up with the infrastructure offerings we have.”
As more mainstream enterprises adopt inferencing, including agentic AI, they will seek vendors that deliver AI operations software and professional services to support use cases and implementation, Kimball said. He expects large enterprises to adopt inferencing this year, with broader commercial enterprises following in 2027.
A Forrester survey conducted toward the end of 2025 found that AI infrastructure skills ranked among the top technology challenges, said Forrester analyst Naveen Chhabra, underscoring a persistent talent gap.
“Who's got the engineering capability to set up those servers?” Chhabra told Data Center Knowledge. “It is a team of highly specialized folks, and the average enterprise doesn't have that skill.”
This skills shortage makes AI operations software (AIOps), which uses AI to automate infrastructure monitoring and management, critical for enterprises, Kimball said. “If I can employ AIOps in my organization to manage my infrastructure, I don't have to worry about my people not being qualified to support that GPU farm I've just deployed,” he said. “I don’t have to worry about them being the bottleneck to standing up AI because I'm relying on automation through AIOps.”
AI also opens the door to vendor switching. With previous technologies such as virtualization and containers, enterprises typically stuck with their incumbent server brand. AI is different. Its unique requirements around performance, cost, power, cooling, and networking are prompting CIOs to evaluate multiple vendors rather than default to their usual choice.
“Whereas before, I would only buy Dell, now I'm evaluating Dell, HPE, and Lenovo to make sure that I'm choosing the right one,” said Kimball.
The vendors that can bridge this gap through professional services and software management tools are positioned to win the enterprise wave. Here’s how Dell, HPE, Lenovo, and Supermicro are each staking their position.
Dell Technologies: Tailored Solutions and AIOps Leadership
Dell tailors its AI server offerings for two distinct customer types: cloud service providers and enterprises.
About 15 years ago, Dell’s Data Center Solutions group provided data center equipment to leading cloud providers such as AWS, Baidu, and Tencent before those hyperscalers shifted to original design manufacturers (ODMs). That engineering know-how now supports Tier 2 CSPs with their AI deployments, said Armando Acosta, Dell’s AI server product planner. He cited Dell’s experience of deploying clusters of 100,000-plus GPUs in 75,000-square-foot data halls, including deployment, integration, and troubleshooting services at that scale.
“Now you have the rise of these Tier 2 CSPs, and they want the same agility, velocity, and engineering experience that we gave 15 years ago,” Acosta said in an interview with DCK.
For enterprises, analysts say Dell has long emphasized ease of use, making its solutions more accessible to mainstream IT organizations. Its management software and professional services organization are differentiators, Acosta said.
Dell’s AI server management tools include:
iDRAC for GPU health monitoring and telemetry;
Integrated Rack Controller for rack-scale visibility, such as leak detection and predictive maintenance on liquid-cooled racks; and
OpenManage for centralized management.
These tools are part of Dell’s broad AIOps solution, which Kimball identifies as the company’s biggest strength. “This is the stuff that really has a big impact on an IT organization,” he said.
Dell Services also advises on use-case prioritization, data management and governance, and model governance – key gaps for mainstream IT teams, Acosta said.
Dell Integrated Rack 7000 (IR7000) series. (Image: Dell Technologies)
Lenovo: Hybrid AI and Liquid-Cooling Expertise
Lenovo’s enterprise strategy focuses on turnkey hardware and software, professional services, and liquid cooling technology. Its services span data preparation and management, network setup, architecture design, data science expertise, and AI ethics advisory.
“AI is new for many of our customers,” Patti said. “Every CIO needs to rapidly have an AI strategy – if they don’t have one. They need the handhold, the guidance, the advice.”
Lenovo brands its approach the Hybrid AI Advantage: customers train in the cloud and increasingly run inference on-premises for data sovereignty and cost control. To support this, Lenovo offers purpose-built inference servers and TruScale, its pay-as-you-go as-a-service model. On the software side, Lenovo’s XClarity platform handles alerting, patching, and system health monitoring. Through its AI Innovators program – a network of about 60 third-party software partners – Lenovo has built an AI Library of vertical-specific applications.
For cooling, Patti said Lenovo’s sixth-generation Neptune technology supports warm-water cooling up to about 45°C, enabling chiller-less operation in suitable climates and reducing energy consumption.
Lenovo has another advantage at the high end of the market. It has the largest footprint among hyperscalers’ OEMs and a strong presence in supercomputing, Kimball said. Through its “ODM Plus” model, Lenovo customizes systems to hyperscalers’ specific requirements.
Lenovo ThinkSystem SC777 V4 Neptune. (Image: Lenovo)
HPE: HPC Heritage and Full-Stack Solutions
HPE draws on its SGI and Cray acquisitions and its own high-performance computing division to provide AI solutions that span supercomputing systems, liquid cooling, system management, and software. “When it comes to full-solution development, deployment, test, and then overall service and maintenance, I don’t think anybody else comes close to the capability that we have,” said Christopher Davidson, vice president of HPE’s HPC and AI Customer Solutions, in an interview with DCK.
HPE’s portfolio spans the AI lifecycle, from large-scale training and fine-tuning to inference – for service providers and enterprises. Offerings include:
HPE Cray Supercomputing GX5000 for HPC and AI workloads;
ProLiant Compute systems for training, tuning, and inference; and
HPE Private Cloud AI for rapid deployment of inference and small-model workloads.
While HPE aligns with Nvidia’s MGX modular reference architectures and ecosystem, it tailors rack designs for specific customer needs. It offers modular data centers, which are shipping container-like units that can be deployed quickly without major facility renovations.
GreenLake provides consumption-based pricing and managed services, Davidson said.
As for software, HPE Morpheus Enterprise Software and OpsRamp (as part of HPE Cloud Ops) deliver cloud automation and AI-powered observability, while HPE Compute Ops Management automates deployment, provisioning, and ongoing management.
Security is another differentiator: HPE’s Integrated Lights-Out controller provides a silicon root of trust with quantum-safe security features from the boot level up, which Kimball calls “the most advanced silicon-based security of the server manufacturers out there.”
HPE Cray Supercomputing GX5000. (Image; HPE)
Supermicro: Modular Design and Manufacturing Agility
Supermicro is leaning on modular engineering and in-house manufacturing to capture AI demand at speed and scale. On the company’s fiscal FY2026 earnings call, CEO Charles Liang said Supermicro’s software and services businesses – including data center planning, design and deployment – are growing rapidly.
Liang highlighted SuperCloud Composer, Supermicro’s data center management software, which he said can manage tens of thousands of racks in real time while monitoring power, cooling, and safety telemetry. The platform also provides CPU and GPU workload orchestration – a critical function for AI data centers.
Analysts said Supermicro’s system-integrator roots and modular design approach allow it to mix and match GPUs, CPUs, storage, and networking to meet specific customer requirements. Combined with its own manufacturing, this has translated into a cost-performance advantage.
Supermicro has built a reputation as one of the first to ship new AI server platforms, Chhabra said. However, he cautioned that speed alone won’t confer a durable lead. “There is significant commoditization in this market," he said. "Being first to market doesn't take you too far.”
Supermicro Data Center Building Block Solutions. (Image: Supermicro)
Playing to Their Strengths
Overall, vendors are leaning into their strengths to stand out in a market still largely defined by Nvidia’s ecosystem, analysts said.
“These vendors approach the AI market with a lot of their history in terms of how they were positioned in the market in mind,” Rutten said. “HPE has a terrific HPC background, Dell has a great usability background, and Lenovo has a lot of experience with both enterprise and hyperscalers. And Supermicro’s modular approach and their own manufacturing have helped them.”

## retained text

No document body was retained for this record — only its metadata was stored.