HPE CEO Antonio Neri told CRN that the acceleration of enterprise agentic AI and AI inferencing is "opening up" the booming AI market opportunity in traditional servers, storage, and networking for channel partners. The comments came after HPE reported cloud and AI revenue up 25 percent to $9 billion in its third fiscal quarter ended July 31. Neri said partners are well positioned because demand is accelerating in traditional infrastructure rather than large-scale training of language models, with on-premise enterprise AI deployments cutting cost per token by as much as 60 percent compared to public cloud alternatives.
HPE's AI system orders reached $2.4 billion, climbing more than 30 percent sequentially, while the company's AI systems backlog increased 14 percent sequentially to a new record high. AI system revenue in the quarter totaled nearly $1.6 billion, with enterprise AI infrastructure demand more than doubling during that period. Server revenue jumped 35.3 percent in the quarter to $6.8 billion compared with $5 billion in the year-ago quarter, while traditional server orders were up 75 percent from the year-ago quarter. Storage business revenue rose 10.2 percent to $1.3 billion compared with $1.17 billion in the year-ago quarter, though storage orders grew at twice that rate. Networking revenue of $2.9 billion was up 10 percent on a normalized basis, but networking orders surged 36 percent, about three-and-a-half times faster than revenue. Networks for AI orders alone reached a new high of $700 million, up triple digits, with cumulative networks for AI orders hitting $2.2 billion and surpassing HPE's Fiscal Year 2026 target.
Neri said "by the end of the decade much of the infrastructure that will be consumed will be AI inferencing," which he described as a position of strength for HPE and partners. During the analyst call, he noted that HPE itself has more than 1,200 use cases in the company, with over 300 in production, signaling that the enterprise has reached an inflection point driven by deployment of agentic AI and AI inferencing. HPE Executive Vice President and CFO Marie Myers pointed to a large retail customer deploying on-premise agentic AI workloads specifically to reduce public cloud AI token costs. One telling sign of growing enterprise demand: a multibillion-dollar HPE server deal closed after the quarter ended with a hyperscaler provider planning to use the servers for their own AI inferencing needs.
The buildout of enterprise agentic AI and AI inferencing favors HPE's traditional server and storage business and its HPE private cloud stack because those enterprise customers don't require massive amounts of GPUs or even CPUs, but rather a "tight" infrastructure that effectively brokers the cost of tokens, according to Neri's comments during the earnings call. The number of tokens on premise is growing very rapidly, which Neri said informs HPE on the durability of AI demand visible in the company's pipeline. Supply constraints have limited HPE's ability to convert orders to revenue, though Neri said he expects supply to become more aligned with order bookings going forward, enabling revenue acceleration even as supply continues to pose constraints. As a result of the strong third-quarter results, HPE raised its guidance for the fourth quarter, Fiscal Year 2026, and Fiscal Year 2027. For Fiscal Year 2026, HPE now expects revenue growth of 34 to 37 percent compared with 29 to 33 percent, with non-GAAP earnings per share of $3.75 to $3.85 compared with $3.35 to $3.45. For Fiscal Year 2027, HPE expects revenue growth of 13 to 17 percent with non-GAAP earnings per share growth of 16 to 20 percent compared with 12 to 16 percent. Neri's advice to partners was direct: "Get aggressive! This is the time to scale AI deployment and inferencing with our customers!" The shift from centralized model training to distributed enterprise inference workloads could reshape competitive dynamics between hyperscalers and traditional infrastructure vendors, forcing partners to balance cloud integration skills with renewed on-premise expertise. Channel organizations that can quantify token economics and translate AI cost models into procurement decisions may gain leverage over those treating enterprise AI as simply another compute refresh cycle.