Under the trend of AI, server delivery is delayed, and Hewlett Packard Enterprise Co.'s performance is lackluster, but the narrative of AI computing power remains hot.

Hewlett Packard Enterprise Co. (HPE.US), focusing on AI computational infrastructure and hybrid cloud platforms, released its performance report and future outlook for the fourth quarter of the 2025 fiscal year ending October 31st on Thursday. The AI server equipment manufacturer's gross profit margin has significantly improved. However, it provided guidance for sales in the current quarter that is lower than the average expectations of Wall Street analysts. This largely indicates that the company has not met the high expectations of the market in terms of AI server sales, primarily because some major AI server transactions have been postponed to the calendar year 2026. Nevertheless, overall orders for AI computational infrastructure remain strong, and future collaborations with AMD and Broadcom Inc. are expected to bring strong incremental growth. Hewlett Packard Enterprise Co. has long been one of NVIDIA Corporation, AMD, and other chip giants' core OEM customers in the AI server field. Like Dell Technologies, Inc. and Super Micro Computer, Inc., the company sells AI GPU/ASIC as complete server cluster solutions to enterprises, governments, and research institutions. Hewlett Packard Enterprise Co.'s HPE Cray supercomputing platform and some ProLiant/AI System product lines target enterprise/cloud-based large AI training and inference clusters, integrating Blackwell GPU or XPU modules into standard rack-level servers or full cab solutions. In a performance statement released on Thursday, Hewlett Packard Enterprise Co. stated that it expects revenue for the current quarter ending in January to range between $9 billion and $9.4 billion, with adjusted earnings per share between $0.57 and $0.61. However, according to institutional compiled data, Wall Street analysts' average revenue expectation is around $9.88 billion, with an adjusted earnings expectation of approximately $0.53 per share. Hewlett Packard Enterprise Co.'s revenue outlook range fell short of analyst expectations. For the 2026 fiscal year, the company reiterated its revenue guidance range of 17% to 22%. Revenue for the fourth quarter of the 2025 fiscal year ending October 31st also fell below the average expectations of analysts. CEO Antonio Neri stated in an interview that some server orders supporting AI workloads were delayed until 2026. He emphasized that a large deal in Europe was delayed due to the unreadiness of an AI data center, and a cooperation agreement with the U.S. government was postponed due to a federal government shutdown. "We feel good about the conversion of that European deal," he said. "These are just timing issues." Financial data shows that Hewlett Packard Enterprise Co.'s revenue for the fourth quarter of the fiscal year increased by 14% year-over-year to $9.7 billion, falling short of the average analyst expectation of approximately $9.9 billion. The company's fourth-quarter adjusted earnings per share reached $0.62, exceeding the average analyst expectation and significantly higher than the $0.04 per share in the same period the previous year. The company's fourth-quarter free cash flow was approximately $1.9 billion, significantly higher than the $420 million in the same period the previous year. In terms of gross profit margin, the company's GAAP gross margin was 33.5%, up 270 basis points from the same period last year and 430 basis points sequentially. Under the Non-GAAP standard, it was 36.4%, up 550 basis points year-over-year and 650 basis points sequentially. Hewlett Packard Enterprise Co.'s Chief Financial Officer, Marie Myers, stated in an analyst earnings call following the release of the performance that demand for AI servers continues to surge, especially from sovereign government agencies and some large cloud computing enterprise customers. "It's worth noting that we expect the demand to remain uneven as some of our largest Sovereign customers have longer lead times for orders that may push shipments of AI server systems into the future," she said during the call. Investors seem to be concerned about the company's performance outlook. After closing at $22.90 on the New York Stock Exchange on Thursday, the stock fell over 9% in after-hours trading due to the company's latest performance and future outlook falling short of market expectations. As of Thursday's close, the stock had accumulated a 6.7% increase this year, underperforming the S&P 500 index and the Nasdaq 100 index. While Hewlett Packard Enterprise Co. is seeing continued growth in demand for its AI servers from global enterprises and some sovereign government agencies, the company is also betting on its networking business to be a key pillar of its future performance expansion. Previously focused on AI servers, the company acquired Juniper Networks Inc. for approximately $13 billion in July. Through cost-cutting measures and selling higher-margin networking equipment after completing the acquisition of Juniper Networks, the company's profit margin has improved. Neri stated: "Our core now is a company centered around servers and networking." The fourth quarter of the 2025 fiscal year was the first complete quarter for Hewlett Packard Enterprise Co. following the acquisition of Juniper Networks. Collaborating with AMD and Broadcom Inc. for strong growth? For Hewlett Packard Enterprise Co., revenue related to AI server sales is undoubtedly closely linked to its AI server series products featuring high-performance AI GPUs from NVIDIA Corporation. However, part of the future performance growth is expected to come from AMD's rack-level AI computational solutions. Hewlett Packard Enterprise Co. is set to be one of the first system suppliers to adopt AMD's "Helios" rack-level AI computational cluster architecture. AMD and Hewlett Packard Enterprise Co. will integrate a specially-made HPE Juniper Networking high-performance expansion switch (scale-up switch) in deep cooperation with ASIC and high-performance network infrastructure leader Broadcom Inc. This large-scale AI computational system aims to simplify the deployment of larger-scale AI computational infrastructure clusters and provide more cost-effective and energy-efficient AMD AI computational cluster solutions beyond NVIDIA Corporation's Blackwell series. The collaboration between AMD, Hewlett Packard Enterprise Co., and Broadcom Inc. around the AMD Helios rack-level AI computational cluster is bundling GPU/CPU/NIC, rack system engineering, and high-end network silicon into an open rack-level AI cluster platform. For cloud computing giants like Microsoft Corporation, Alphabet Inc., and Amazon.com, Inc., as well as sovereign AI builders who are investing heavily in AI computational infrastructure, this is not only a high-end alternative to NVIDIA Corporation's vertically integrated solutions but also a foundational template for the integration of "sovereign AI + cutting-edge HPC"where data centers deploy standard building blocks of AI computational power at the rack level that can simultaneously run HPC and large parameter AI models. Hewlett Packard Enterprise Co. CEO Antonio Neri stated: "With the introduction of the new AMD 'Helios' and our specially-designed HPE rack-level networking solution, we are providing significant assistance to cloud computing service providers and large customers in deploying AI applications globally, making rack AI systems more flexible and reducing computational risks, thereby supporting them in scaling AI computational clusters in their businesses on a large scale."