HPE Uses Networking Strength To Offset Memory Shock And Lift 2026 Outlook

A sharp turn in the cycle

Hewlett Packard Enterprise opened its 2026 fiscal year with a statement quarter that said as much about the changing economics of IT hardware as it did about the company’s own transformation.

On the surface, the numbers were straightforwardly robust. Revenue in the January quarter rose 18 per cent to $9.3bn, toward the top end of guidance. Non‑GAAP earnings per share hit a record $0.65, comfortably above the company’s own range, while free cash flow came in at $708m in what is usually a seasonal trough. Operating margin reached 12.7 per cent, underpinned by a 36.6 per cent gross margin.

Underneath, though, this was an earnings print dominated by scarcity. Chief executive Antonio Neri and chief financial officer Marie Myers painted a picture of an industry entering what Neri called a “sharp acceleration in supply tightness” in DRAM and NAND. Triple‑digit price increases between the December and March quarters, and the prospect of elevated costs “well into 2027”, are already reshaping how HPE prices, configures and even selects which orders to fulfil.

In that environment, the company’s post‑Juniper profile is starting to look structurally different. Networking – once an “edge” side‑show inside a server‑centric group – is now almost a third of revenue and more than half of operating profit. Cloud & AI, the other new segment, is being managed with an explicit tilt toward margin over volume.

Networking becomes the profit engine

The headline shift in HPE’s portfolio is clearest in Networking, which now folds in Juniper alongside the former Aruba Intelligent Edge business. In its first quarter under the new structure, the segment delivered $2.7bn of revenue, up 152 per cent on a reported basis and 7 per cent on a Juniper‑normalized view.

Orders grew faster than revenue, with particular strength in the parts of the network that matter most for artificial intelligence workloads. Normalized data centre switching orders were up in the mid‑40s per cent range; routing orders rose in the mid‑20s. In product‑category terms, data centre networking revenue grew 31 per cent and routing 10 per cent on a normalized basis, while campus and branch ticked up 2 per cent and security slipped 5 per cent.

Crucially for investors, Networking is not only growing; it is highly profitable. Segment operating margin came in at 23.7 per cent, slightly ahead of guidance, supported by scale, tight pricing discipline and what Myers called “early Juniper synergies”. HPE has completed “sales day 1”, merging Juniper and Aruba salesforces into a single organization under harmonized compensation plans, and is now turning to deeper integration in supply chain, real estate and marketing.

The company is leaning into a theme it brands “networks for AI” – the fabrics and routers that link AI data centres, sovereign AI facilities and service provider cores. On the back of strong Q1 orders, HPE raised its target for cumulative networks‑for‑AI orders to $1.7bn–$1.9bn by the end of fiscal 2026, up from a prior $1.5bn. That figure is driven by demand from both service providers – whose networking revenue grew 20 per cent on a normalized basis – and large enterprises, which were up a modest 2 per cent.

Neri highlighted the ramp of WiFi 7 as an early indicator of campus demand. Shipments of WiFi 7 access points rose more than ten‑fold, and devices connected to the Mist and Aruba Central cloud platforms increased 28 per cent. In data centres, he pointed to the new high‑density PTX series routers as emblematic of Juniper‑led innovation, citing the ability of one compact chassis to support 16m concurrent sessions – enough, as he put it, to stream a Netflix movie simultaneously to all residents of London and New York.

Symbolically, HPE used the Milano Cortina 2026 Winter Olympic Games as a showcase, running connectivity and security across distributed venues with a slim, AI‑assisted operations team. For investors, the more important point was not the brand halo, but the message that the combined Juniper–Aruba stack can now address campus, branch, data centre and AI interconnect with a single, software‑heavy portfolio.

On the back of this performance, management raised its full‑year Networking revenue growth outlook to 68–73 per cent on a reported basis, implying mid‑ to high‑single‑digit growth on a normalized comparison. The operating margin range for the year stays at the “low 20s”, a rather cautious stance after a first quarter close to 24 per cent. Myers argued that, with only the first phase of Juniper integration complete, maintaining the range is prudent; the implication is that any upside from cost synergies or mix will drop through rather than be pre‑committed.

Cloud & AI: growth sacrificed to margin

If Networking is the visible winner from the new supply landscape, Cloud & AI is the segment where the trade‑offs are most evident. The unit, which houses traditional servers, AI Systems, storage, hybrid cloud software and HPE Financial Services, reported $6.3bn of revenue, down 3 per cent year‑on‑year in line with previous guidance.

Despite the revenue decline, Cloud & AI’s operating margin expanded to 10.2 per cent. Operating profit grew 18 per cent versus a year earlier and 4 per cent sequentially, helped by price increases, disciplined cost control and a conscious decision to “pivot to higher‑margin, more profitable components of our business”, as Myers described it.

Server revenue fell 3 per cent, reflecting the expected timing of AI server shipments. Traditional servers saw strong average unit price growth and demand, as customers both modernized on‑premise infrastructure and pulled forward orders to get ahead of component inflation. But AI Systems revenue was light in the quarter. HPE nevertheless booked $1.2bn of AI Systems orders and ended Q1 with a record $5bn AI backlog, largely from enterprise and sovereign customers rather than hyperscalers.

The mix of orders has shifted in favour of those high‑margin categories. Since early 2023, the proportion of AI orders coming from enterprise and sovereign clients has risen, consistent with HPE’s strategy of prioritizing profitability and attachment opportunities over lower‑margin cloud titan business. Management again flagged that AI revenues will be “uneven” through the year, with some large sovereign projects carrying extended lead times. Most AI shipments are still expected in the second half, with the third quarter flagged as the largest AI revenue period and profitability weighted to the fourth quarter.

Storage, which bundles hardware, private cloud and GreenLake software, eked out 1 per cent revenue growth but showed stronger underlying dynamics in HPE’s own intellectual property. Alletra MP – the company’s next‑generation storage platform – delivered 42 per cent order growth and strong double‑digit revenue growth, marking a fifth straight quarter of double‑digit expansion. HPE is in the process of exiting lower‑margin third‑party, non‑IP storage, which dampens top‑line optics but lifts mix.

Financial Services posted flat revenue but a record 27 per cent return on equity. In a world of constrained component supply and rising prices, Neri framed HPE FS as a strategic lever: able to finance GreenLake as‑a‑service deployments, extend the life of existing kit and provide certified pre‑owned hardware where new supply is tight. Management expects networking assets to form a growing share of the FS book now that Juniper’s portfolio is in‑house.

The net effect of these currents is a more cautious growth profile for the segment. HPE trimmed its full‑year Cloud & AI revenue outlook to mid‑ to high‑single‑digit growth, from an earlier mid‑single‑digit to low‑double‑digit range. The non‑GAAP operating margin target remains 7–9 per cent for the year, with Q2 expected around the midpoint and margins fluctuating thereafter with AI shipment timing and memory costs.

Navigating a memory shock

What ties the segments together is a pricing and procurement regime that looks very different from the server cycles of the past decade.

DRAM and NAND now account for more than half of the bill of materials of a typical server and an increasing share of storage systems. Neri told investors that between the December and March quarters the industry has already seen triple‑digit increases in memory prices, with further double‑digit rises expected “as we go forward”. He does not believe supply can meet all of the demand HPE sees in its order books, even after signing expanded multi‑year agreements with key silicon and memory suppliers.

Faced with that, HPE is prioritizing three things. First, it is securing as much supply as possible and is willing to hold more inventory – which ended the quarter at $6.9bn, down year‑on‑year but up sequentially – and higher purchase commitments to guarantee future shipments. Second, it is “protecting our margins”, in Myers’ formulation, by systematically passing through component inflation. The company has implemented DRAM‑related price increases since November, shortened quote commitment windows and rewritten terms to allow repricing of orders between quotation and shipment if commodities move.

Third, it is steering demand, particularly in the enterprise. Sales, supply chain and pricing teams are now closely coordinated to offer alternative configurations, including lower memory footprints where appropriate, and to favour higher‑margin orders. Networking, which uses much less memory, sits naturally at the top of that priority list. Traditional server orders with attractive service attachments also score highly; low‑margin, memory‑heavy configurations or more commoditised AI deals are deprioritised.

Interestingly, management insists that so far it has seen “0 impact on demand” from these measures. Customers, Neri said, are far more concerned with getting product quickly than with headline pricing. In practical terms, many are accepting higher average unit prices but adjusting configurations to stay within budget and secure earlier delivery.

That calculus feeds through directly into the outlook. HPE raised its full‑year non‑GAAP EPS guidance by $0.05 to a range of $2.30–$2.50. GAAP EPS guidance increased by $0.40 to $1.02–$1.22, helped by lower net interest expectations, with other income and expense now forecast at $540m–$590m versus $650m previously. Free cash flow guidance moved up as well: management now expects at least $2bn, from a prior $1.7bn–$2bn range.

At the same time, Myers warned that investors should not lean on historical seasonality. The combination of Juniper’s integration, AI backlog timing and component volatility means that familiar patterns – such as a typical sequential revenue decline from January to April – are unlikely to hold. For the second quarter, HPE is guiding revenue to $9.6bn–$10bn, implying 5 per cent sequential growth, with Networking expected to grow 142–152 per cent year‑on‑year reported and at the high end of its normalized range. Non‑GAAP EPS is guided to $0.51–$0.55, with operating margin expected to be lower than Q1 as annual compensation resets, marketing spending and higher commodity costs bite.

GreenLake, AI and the long game

Away from the immediate supply‑chain firefighting, HPE also used its prepared remarks to underline the progress of its longer‑term strategic pivots.

GreenLake, its hybrid cloud and as‑a‑service platform, is now approaching 50,000 customers. Annualized recurring revenue is on track to reach $3.5bn by the end of fiscal 2026, driven not just by infrastructure consumption but by subscription services across networking, storage and cloud software. Newer offerings such as VM Essentials – pitched as a lower‑cost alternative in a world of “escalating cost of legacy virtualization software” – have delivered high double‑digit growth in new logos for three consecutive quarters.

Private cloud AI orders have increased for four quarters in a row, with wins across both enterprises and service providers. Neri spoke of European industrial groups and utilities, such as Siemens Energy, adopting “agentic AI” workflows on‑premise, often combining HPE compute, networking and GreenLake software.

Internally, the company’s “Catalyst” programme – a broad modernization and cost‑efficiency effort – is leaning heavily on AI to extract savings. HPE cited generative AI tools that can cut engineers’ search time for technical insights by 90 per cent and configuration recommendation engines that may shorten quote cycles by 30 per cent. These projects, alongside Juniper synergies, are a visible component of HPE’s commitment to reach at least $3 in non‑GAAP EPS and more than $3.5bn in free cash flow by fiscal 2028.

There are still uncertainties to watch. Management mentioned the recent US Supreme Court tariff decision as a risk that needs “greater clarity” before its impact can be fully assessed. The company is also monitoring operations in the Middle East closely amid geopolitical tension, and continuing to work through its previously announced H3C transactions, which are on track to close in the first half of calendar 2026.

For now, though, HPE is presenting itself as one of the relative winners from a difficult hardware cycle: a company that has used an acquisitive pivot into networking, a disciplined stance on pricing and a sharpening focus on profitable AI and hybrid cloud niches to turn a memory shock into an opportunity to upgrade its own earnings profile. Investors will now look to the back half of the year – and the conversion of that $5bn AI backlog – to see whether the new shape of the portfolio can hold its margin promises once the revenue mix shifts back toward silicon‑heavy compute.