Hewlett Packard Enterprise

Hewlett Packard Enterprise reported exceptional fiscal third-quarter results, as revenue increased 34% year over year while non-GAAP earnings per share exceeded the high end of guidance.

Why it matters: HPE is showing momentum across its portfolio, driven by artificial intelligence-related demand, reinforcing our view that AI is creating a broader and more durable enterprise infrastructure cycle. Hybrid cloud, general-purpose servers, and AI servers all saw solid growth. This matches a similar pattern to peer Dell, as both are seeing AI demand feed into multiple parts of their product portfolio. The AI systems backlog rose again in the quarter even as AI systems revenue was up 73% sequentially. Networking revenue also did well, up 10% on a normalized basis even as normalized orders rose 36%. The combined Juniper/HPE portfolio is seeing a nice proof point with the latest Oracle deal, which will combine Juniper switches with other hardware in a multigigawatt buildout.

The bottom line: We are raising our fair value estimate for no-moat HPE to $66 from $64 as we incorporate higher revenue expectations in both 2026 and 2027. Overall, we thought earnings were solid and are a bit surprised by the after-hours selloff and view shares as undervalued. HPE raised fiscal 2026 guidance across the board, and also provided an updated 2027 outlook that was above its prior 2027 outlook. We adjusted our fiscal 2027 EPS estimate to $4.60 compared with $3.88 before. HPE also closed a $3.5 billion deal with an unnamed hyperscaler customer after the quarter closed—not currently included in the backlog number—and the AMD Helios opportunity could also provide further upside.

Bulls say: Overall, we thought results were quite strong, reinforcing our more bullish stance on HPE. Networking is showing strength, margins are doing well, and AI demand is driving growth. Eventually, we think this will lead to a higher share price.

Our fair value estimate is $66 per share. Our valuation implies an adjusted price/earnings multiple of 18 for fiscal 2026 earnings, growing into a forward P/E multiple of 13 times by 2029 (on a rolling fair value estimate basis).

Our overall thesis on HPE encompasses durable sales in the AI-optimized server market, robust growth in the general-purpose server market driven by agentic AI, enhanced margins in the hybrid cloud business, and improved growth rates in both the hybrid cloud and networking segments. We expect AI demand and the now-closed Juniper acquisition to be the two key factors driving valuation of the shares. As the company gets more credit for growth and its higher-margin networking business, we expect shares to rerate higher.

In the AI-optimized server market, we forecast HPE achieving approximately $12 billion in business by 2028, with solid mid- to high-single-digit growth after that. We expect GP servers will compound at 19% over the next three years. This will help HPE’s server business grow at a compound annual growth rate of 17% over the next five years. We view AI-optimized servers as a key revenue growth driver for the next several years. The margin profile of this revenue growth is a key debate. We assume the cloud & AI segment can maintain a 12% operating margin over time, and the margin issues of early 2025 are indeed now looking temporary, as expected. We also forecast a more steady performance in the hybrid cloud business, with annual growth rates improving to the midsingle digits (from negative growth prior to 2025).

Networking is becoming an increasing focus for HPE, with the Juniper acquisition. We forecast networking will grow at a 31% CAGR through 2028, with normalized margins of 27%. We expect the networking business and a healthy server business will enable HPE to reach a 37% gross margin over time compared with low-30s recently.

Overall, including the Juniper acquisition, we forecast a five-year revenue CAGR of 14%, with gross margins improving as networking becomes a larger part of the business, while operating margins expand to 13% by 2030, driven by a higher percentage of the business coming from networking, execution of $1 billion in expense synergies, a higher percentages of storage sales coming from IP heavy products, and a turnaround from previous server-related discounting.

We don’t believe Hewlett Packard Enterprise has an economic moat. HPE’s returns on invested capital are decent for a competitive, commoditized industry at roughly 13%-14% (including goodwill), but the spread is small enough and we believe future returns will depend more on execution than structural competitive advantages.

HPE’s largest businesses are servers and storage, where vendors primarily compete on price with largely undifferentiated products. Networking offers greater differentiation and higher margins, but HPE has historically been a second-tier player behind leaders such as Cisco and Arista. We also do not view AI-optimized servers as moatworthy. Despite strong growth, current margins are no better than general-purpose servers, and we expect similar competitive dynamics.

HPE is an original equipment manufacturer, selling finished products manufactured by original design manufacturers under the HPE brand, primarily to enterprises. Competitors such as Dell, Lenovo, and Supermicro generally offer similar products, while original design manufacturers such as Inventec, Wistron, and Foxconn manufacture hardware to original equipment manufacturer specifications. Because all OEMs have access to similar supply chains and differentiated components tend to accrue value to their manufacturers, such as Nvidia in AI servers, hardware differentiation is limited. We therefore do not see material intangible assets. OEMs mainly compete on price and service.

OEMs nonetheless add value by coordinating complex supply chains, providing enterprise support and services, and maintaining brands and salesforces capable of serving a fragmented global enterprise market. This explains why OEMs remain relevant with enterprises. However, hyperscalers such as Alphabet, Microsoft, Amazon, and Meta have the scale and expertise to buy directly from ODMs. As enterprise computing has shifted toward public cloud, this has expanded the portion of the market served directly by ODMs while traditional enterprise server demand has stagnated, creating a structural headwind for OEMs.

AI-optimized servers are a new source of server growth, but we view this as a revenue opportunity rather than an improvement in industry economics. Hyperscalers remain major buyers, benefiting ODMs, while OEMs are also seeing stronger demand. Yet AI servers have often carried lower margins than general-purpose servers, and we see little reason OEMs should capture more value from them given similar competitive forces.

Storage has broadly similar economics. HPE competes with Dell, NetApp, IBM, and Pure Storage in storage arrays integrated into broader IT stacks. Much of the value accrues to component suppliers of NAND, DRAM, and hard drives, while OEMs again primarily serve enterprises and ODMs can more efficiently serve hyperscalers. Storage offers somewhat more software differentiation through management tools and integration, supporting better margins than servers. Still, refresh cycles of roughly three to five years force vendors to repeatedly compete on price, and customers generally demand interoperability rather than closed systems. Market-share shifts, including gains by Pure Storage, reinforce our view that execution matters more than durable competitive advantages. New architectures such as hyperconverged infrastructure or consumption-based pricing do not materially change these economics because competitors can offer similar models.

Networking is HPE’s strongest segment and the closest to being moatworthy. Networking equipment and software can support higher margins and stronger returns, and leaders such as Cisco and Arista possess wide moats. However, HPE lacks their scale and market leadership. Cisco holds dominant positions across many categories and offers a more complete end-to-end solution, while Arista has differentiated through high-speed networking. HPE has meaningful wireless access-point share but remains well behind Cisco, and enterprise switching is highly competitive.

Networking also has switching costs because customers can benefit from using a common vendor, but these are uneven. Enterprises can mix vendors across locations and product categories, and some networking products are easier to replace or combine than others. As a result, HPE’s current position does not demonstrate enough pricing power or differentiation for a moat.

The Juniper acquisition should materially strengthen HPE’s networking portfolio, particularly in data center switching and routing, and networking margins are attractive in the mid-20s. We therefore view networking as HPE’s best potential source of a moat. If HPE can gain share, maintain margins, and successfully integrate Juniper, we could consider the segment narrow-moat worthy. For now, however, the company overall remains no-moat.

If HPE can execute on its Juniper acquisition, leading to higher margins, and reinvigorate growth in its hybrid cloud business while benefiting from the AI server wave, there should be upside.

HPE is one of the few companies to have a portfolio that spans servers, storage, and networking. To the extent that this completeness matters more over time, it should benefit HPE.

The AI growth wave is just starting, and HPE will be a key beneficiary as inference workloads for enterprises begin to take off.

The AI server market’s current growth may not be durable, and future growth will depend on AI demand proving maintainable. OEMs like HPE will struggle to see this structurally change the profitability of their overall server business.

The general-purpose server and storage markets are seeing stagnant growth as enterprises shift more workloads to the public cloud. This limits the structural growth opportunities for OEMs like HPE.

HPE does not have a strong history of executing acquisitions effectively. The purchase of Juniper may not be executed well.