Skip to main content
Back to Blog
EnterpriseAI/MLCloud Computing
7 September 20267 min readUpdated 8 September 2026

HPE Reports Strong Growth Despite Rising Datacenter Equipment Costs

HPE reports stronger growth across servers, networking, and AI systems The generative AI boom is increasing demand across much of the datacenter equipment market. Dell has becom...

By Hardware Team

HPE reports stronger growth across servers, networking, and AI systems

The generative AI boom is increasing demand across much of the datacenter equipment market. Dell has become the world’s largest original equipment manufacturer, a position once held by the original Hewlett Packard after its 2001 acquisition of Compaq. Compaq had previously acquired Tandem and Digital Equipment Corp.

Hewlett Packard Enterprise remains among the top five server manufacturers and continues to serve millions of companies worldwide. Its broader position across servers, storage, networking, support, and financing may matter more over the long term than winning a small number of multibillion-dollar AI system contracts. HPE also has limited access to GPU allocations from Nvidia and AMD, constraining how quickly it can expand its AI systems business.

HPE’s financial results for the third quarter of fiscal 2026 indicate that its business is improving across enterprises, sovereign organizations, service providers, and smaller cloud providers.

Third-quarter financial results

For the quarter ended in July, HPE reported revenue of $12.21 billion, up 33.7 percent year over year and 14.4 percent sequentially. Cost controls, a richer product mix, and price increases passed through on CPUs, GPUs, DRAM, flash, and networking contributed to the improvement.

Operating income increased 5.6 times to $1.39 billion, while net income rose 5.5 times to $1.51 billion. Net income represented 12.4 percent of revenue. The results were among the strongest quarters in HPE’s datacenter history, with comparable systems profitability last seen during the Unix systems boom of the late 1980s and early 1990s.

HPE ended the quarter with $6.22 billion in cash and investments, an 11.6 percent increase from the same period a year earlier.

Cloud & AI remains HPE’s largest group

HPE’s Cloud & AI group combines servers, storage, financial services, and technical support for those products. Its Networking group is built around the acquired Juniper Networks business and also includes HPE Ethernet switches, including the Slingshot high-performance computing networking technology developed by HPE’s Cray unit.

Cloud & AI revenue rose 25.4 percent to slightly more than $9 billion, including a 17.3 percent sequential increase. The results show HPE converting both AI and traditional system orders into revenue while its backlog continues to grow.

Cloud & AI operating income increased more than threefold to $1.54 billion. Within the group:

  • Server revenue increased 35.3 percent to $6.77 billion.
  • Storage revenue increased 10.2 percent to $1.29 billion.
  • Financial services revenue remained steady at $883 million.

HPE does not report operating income for these individual divisions.

Networking benefits from Juniper integration

The Networking group generated $2.89 billion in third-quarter revenue, up 74.9 percent year over year. HPE has been attaching more networking products to server and storage sales, one of the strategic reasons for acquiring Juniper Networks.

Campus and branch switching remained the group’s largest business, generating $1.44 billion, up 31 percent year over year and representing just under half of networking revenue. Routing revenue reached $788 million, an increase of 3.7 times from the prior-year quarter. Hyperscalers, cloud builders, and smaller cloud providers also purchase routing equipment to connect their datacenters.

One such customer is Oracle, which operates its own cloud. After the close of the third quarter, HPE announced an agreement to supply Oracle with routers and switches. The deal primarily involves Juniper QFX switches based on Broadcom’s 102.4 Tb/sec Tomahawk 6 switch ASIC, which launched in June of the previous year and is now shipping. The agreement also includes Juniper’s AIOps software for network management.

HPE said the arrangement is part of a potential multigigawatt datacenter rollout that could extend across multiple years. HPE granted Oracle warrants for up to 4.2 million HPE shares. Before the announcement, those shares would have been worth more than $200 million. HPE did not disclose the actual vesting terms for the warrants.

The Oracle agreement is separate from a $3.5 billion deal HPE signed with an unidentified hyperscaler. Antonio Neri, HPE’s chief executive officer, described that customer as a hyperscaler using HPE infrastructure for internal AI inferencing rather than for operating a public cloud service. He said the arrangement involves multibillion-dollar AI inferencing infrastructure for internal enterprise use.

AI system orders and backlog

HPE recorded just under $1.6 billion in AI server revenue during the third quarter of fiscal 2026, approximately 7 percent higher than in the same period a year earlier.

The company added $2.4 billion in new AI system orders during the quarter, while its AI server backlog reached $6.8 billion at quarter end. HPE’s figures appear to include high-performance computing systems in this category, as many current HPC deployments also involve AI workloads.

HPE separately reported $700 million in AI networking orders and an AI networking backlog of $800 million. It did not disclose AI-related networking revenue, but said AI networking orders were up 3.5 times year over year.

Traditional server demand also accelerated

The non-AI server business produced $5.17 billion in sales during the quarter, up 52 percent year over year. HPE, like Dell, is benefiting from an upgrade cycle among customers replacing older ProLiant servers.

Newer replacement systems typically contain more CPUs, DRAM, flash storage, and hard drives. These higher-density systems cost more, although their density and power efficiency can justify the expense compared with maintaining racks of older equipment. Component prices have also continued to rise across the system.

Core datacenter revenue reaches another record

HPE’s datacenter business predates the generative AI boom and the tenure of its current leadership. Generative AI began contributing to revenue growth in fiscal 2024, while the addition of Juniper accelerated the trend.

HPE has now set records for core datacenter revenue for five consecutive quarters. This measure includes servers, storage, networking, technical support, and financing. HPE remains smaller than Dell, Supermicro, Inspur (IEIT), and Lenovo in server sales, but the difference is narrower when networking and financing are included.

In the third quarter, HPE’s core systems revenue reached $10.49 billion, up 35.1 percent year over year and 15.6 percent sequentially. An allocation of operating income indicates that the core systems business generated approximately $1.86 billion at the operating level.

Future revenue growth will depend partly on the GPU allocations HPE and its customers can obtain. It will also depend on whether customers can secure sufficient datacenter power and facilities to deploy those systems.

Fourth-quarter outlook

For its fiscal fourth quarter, HPE expects revenue between $13.9 billion and $14.8 billion. The company expects Networking revenue to grow between 11 percent and 13 percent, while Cloud & AI revenue is projected to increase between 60 percent and 72 percent.

The outlook remains tied to the availability of power and facilities needed to support allocated GPUs.

Neri said customers are reallocating and increasing their technology budgets at levels not seen since the dot-com boom. He also said customers initially reacted cautiously to the cost of AI infrastructure, studying market trends and spot pricing before adjusting their plans.

According to Neri, customers now recognize that delaying AI deployments is not an option. They are focusing more closely on budget allocation and token economics. He said on-premises deployments can provide cost benefits of up to 60 percent per token in certain cases, and that he is not currently seeing significant customer hesitation.