HPE earnings offered a useful read on how enterprise AI spending is moving through more of the technology stack. Hewlett Packard Enterprise reported record fiscal third-quarter revenue of $12.2 billion on September 2, up 34% from a year earlier, and raised its fiscal 2026 outlook. The result matters less as a single-company scorecard than as evidence that AI investment is supporting demand for servers and networking as well as the chips that have dominated the conversation.
The company reported results for the quarter ended July 31. On a GAAP basis, diluted earnings per share were $1.06; non-GAAP diluted earnings per share were $1.11. Revenue was $12.213 billion in the detailed financial tables, which HPE rounded to $12.2 billion in its headline results. Its September 2 Form 8-K confirms that the company furnished the earnings release that day.
HPE earnings point to a broader enterprise AI build-out
Two business lines make the pattern especially clear. HPE said its Cloud & AI segment generated $9.0 billion in quarterly revenue, up 25.4% year over year. Within that segment, server revenue was $6.8 billion, up 35.3%. Its Networking segment reached $2.9 billion, up 74.9%, with data-center networking revenue up 112.2% and routing revenue up 270.0%.
Those are not interchangeable figures, and they should not be treated as a precise measure of the AI market. HPE sells across a broader enterprise-technology portfolio, while its reported segments include products and services that are not exclusively tied to AI. Still, the composition is instructive: a company positioned in both compute and networking is seeing strength in both. That is consistent with an enterprise build-out that involves connecting, operating and scaling AI workloads, not simply acquiring processors.
HPE raised its fiscal 2026 revenue-growth outlook to 34% to 37% and its non-GAAP diluted-EPS outlook to $3.75 to $3.85. It also lifted its fiscal 2027 revenue-growth framework to 13% to 17%. Forecasts are management guidance, not completed results, and HPE's regulatory filing explicitly identifies supply-chain dynamics, component availability, trade policies and the integration of Juniper Networks among the factors that could make actual outcomes differ.
The constraint is still physical
The independent market context is less celebratory. Reuters reported that HPE shares fell more than 3% in extended trading after the results, with persistent supply constraints weighing on the reaction. In an interview cited by Reuters, Chief Financial Officer Marie Myers identified memory as the main bottleneck, followed by NAND, CPUs and drives. HPE's inventory stood at $11.82 billion at the end of July, compared with $7.16 billion a year earlier, a change Reuters attributed to higher commodity costs and targeted purchases supporting orders and backlog.
That qualification is central. A high revenue-growth rate can coexist with a harder procurement environment, and rapid demand does not automatically turn into smooth deployment. For enterprises, the practical questions remain lead times, system configuration, power and network capacity, and the cost of assembling complete infrastructure. For HPE, the raised outlook is a statement of confidence; it is not a guarantee that those constraints will fade.
Why this matters beyond HPE
Recent AI-infrastructure coverage has understandably focused on the largest chip suppliers and headline data-center commitments. HPE's report offers a different lens: the network and server layers may be becoming a more visible part of the spending cycle. Networking was HPE's fastest-growing reported segment in the quarter, while Cloud & AI produced the larger revenue base. That combination suggests that the business case for enterprise AI increasingly includes the systems around the model, from computing and storage to data-center connectivity.
It also sharpens the distinction between announced AI ambition and operating reality. HPE's growth and guidance point to genuine customer demand across its portfolio; the supply discussion shows the limits of treating demand as the whole story. Readers watching enterprise AI adoption should follow whether future results sustain both compute and networking growth, whether component constraints ease, and how much of the demand reflects durable production workloads rather than early capacity reservation.
For now, the takeaway is measured. HPE's quarter adds evidence that AI infrastructure demand is broadening beyond a single category. It also shows that the next phase of the build-out remains dependent on the less glamorous work of sourcing components and delivering complete systems.
