Hewlett Packard Enterprise Company (NYSE: HPE) — Dear Against Itself: A Real Networking Transformation Riding a Memory Windfall

Report date: 2026-06-11 · Fiscal year ends October 31 Price (2026-06-10): $45.49 · Shares (diluted): ~1.324B · Market cap: ~$60B · Enterprise value: ~$76–78B

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information only — not investment advice and not a recommendation to buy or sell any security. The analytical body that follows takes no position and sets no price target; the only opinion expressed anywhere is in this clearly-labeled block.

Verdict: HOLD / accumulate-on-weakness. Great asset hidden in a mediocre conglomerate, at a fair-to-full price. Fair-value zone ~$33–$45; I would add below ~$38 and trim above ~$50. Not a short.

For nine years after the 2015 spin, HPE was the market’s archetypal value trap — a sub-scale box-maker trading at 5–8x earnings because it deserved to. In 2025–26 the market finally re-rated it ~2.5x off its April-2025 lows to $45, and — this is the uncomfortable part for bears — the re-rating is largely earned. The ~$14B Juniper acquisition genuinely transformed the mix: Networking is now ~30% of revenue but >50% of operating profit at 23%+ margins, giving HPE its first defensible franchise of scale. Forward numbers have inflected hard — FY26 non-GAAP EPS guidance was raised >40% to ~$3.40 and free cash flow to ≥$3.5B, putting the stock at just ~13x forward earnings and a ~5.8% forward FCF yield. On those numbers HPE is the cheapest name in its peer set. So why only a HOLD? Because three things sit underneath the print that the bulls are waving through. First, the quality of the forward jump is mixed: a large slice of the FY26 server surge is a DRAM/NAND memory super-cycle inflating ASPs (server revenue +33% on price, not units, which were supply-constrained) — a pass-through windfall that reverses when memory normalizes. Second, the growth engine is the lowest-quality pool: AI servers carry ~single-digit margins (Nvidia keeps the economics), and HPE is the global #5 server vendor at ~3% share and a share-losing #5 in storage that just wrote off $1.6B of goodwill. Third, the price is dear against its own history — the 97th percentile of HPE’s own 10-year valuation, with the sell-side mean target ($30) sitting 34% below spot, zero insider open-market buying, negative tangible book, a pro-cyclical buyback that got paused at the bottom, and a 5.7–7% preferred-conversion dilution overhang in 2027.

The honest framing is “cheap against peers, dear against itself.” You are buying a genuinely better company than the 2015–2024 HPE, but you are paying for the transformation and a cyclical tailwind at a decade-high relative valuation, with Elliott’s catalysts (deleveraging, 75%-of-FCF return, a possible networking separation) now largely known and partly priced. That is a HOLD, not a chase. Framing: a re-rated value stock that has caught up to — and slightly outrun — its own fundamentals. Conviction: medium. Flips bullish if, after the memory cycle normalizes, server margins hold double-digits and Networking sustainably runs >50% of operating profit at mid-to-high-20s margins (proving the mix-shift is structural, not a windfall) — or if the Strategy Committee separates Networking and the market pays a Cisco/Arista-type multiple for it. Flips bearish if the memory cycle rolls over and server margins revert toward 7–8% while AI-backlog conversion slips and storage takes a further write-down — making the FY27 framework (EPS +12–16%, FCF ≥$4.5B) unreachable. Tag: the value trap finally sprung — now priced like it.

1. Executive Summary

Hewlett Packard Enterprise is a $34B-revenue (FY2025) enterprise IT infrastructure vendor spun out of the old Hewlett-Packard in November 2015. It sells general-purpose and AI servers (the largest segment), enterprise networking (Aruba plus, since July 2025, Juniper), hybrid-cloud/storage, and a captive financing arm. The investment debate is not about whether HPE is a good business in the abstract — large parts of it are structurally mediocre — but about whether a genuine, acquisition-driven improvement in its earnings mix justifies a stock that has re-rated to the top of its own decade-long valuation range.

What changed. Two things broke HPE out of its perennial value-trap range. (1) The ~$14B all-cash Juniper Networks acquisition (closed 2 July 2025) roughly doubled HPE’s networking franchise, the only part of the company that earns software-like margins (23–25% operating margin), and made Networking the largest profit contributor despite being only ~30% of revenue. (2) A confluence of AI-server demand and a DRAM/NAND memory super-cycle drove the Server segment’s revenue up double digits, with FY2026 guidance raised aggressively — non-GAAP EPS to ~$3.40 (a >40% in-year raise) and free cash flow to ≥$3.5B — pulling forward HPE’s own long-term targets by two years. Activist Elliott Management (>$1.5B stake) added a catalyst overlay: a cooperation agreement, a new board-level Strategy Committee, accelerated deleveraging, and a commitment to return ~75% of free cash flow from FY2027.

What hasn’t changed. Underneath the inflection, HPE remains the #5 global server vendor (~3% share) in a market where ~47% of units bypass OEMs entirely (ODM-direct), where AI servers are low-margin GPU pass-through, and where its server operating margin still fell from 11.4% to 7.6% in FY2025 even as revenue grew. Its storage business (Hybrid Cloud) is a share-losing #5 that absorbed a $1.6B goodwill impairment in FY2025 — management’s own admission that the franchise has no excess earning power. GAAP FY2025 was a net loss to common shareholders (−$59M) after preferred dividends. Free cash flow fell 57% to $986M. The balance sheet was levered up to fund Juniper (cash drained from $14.8B to $5.8B; total debt ~$21.5B), tangible book is deeply negative (goodwill + intangibles of $30.1B exceed equity of $24.7B), and a $1.5B mandatory-convertible preferred will dilute common holders by ~6% in 2027.

The tension. On forward numbers HPE looks cheap — ~13x earnings, ~5.8% FCF yield, the lowest EV/EBITDA in its peer group. On its own history it is dear (97th percentile composite valuation). The reconciliation is that forward estimates have genuinely rocketed, but a meaningful share of the jump rides a cyclical memory windfall and full-year Juniper consolidation rather than durable, organic, high-quality earnings power. The market is paying today for a transformation that is real but only partly proven, and capitalizing a cyclical tailwind as if it were structural. This memo argues both sides on the evidence and leaves the recommendation to the labeled block above.

2. Business Overview

HPE sells the hardware, software, networking, and financing that enterprises, governments, and service providers use to run on-premises and hybrid IT. In FY2025 it reported $34.3B of net revenue across five segments (realigned for FY2026, discussed below). Using the FY2025 reporting structure:

(Source: HPE FY2025 Form 10-K, segment footnote. Consolidated net revenue is after intersegment elimination of $333M.)

Server (~51% of revenue, ~37% of profit). General-purpose ProLiant x86 servers, workload-optimized and AI/accelerated systems (including the Cray EX supercomputing line and direct-liquid-cooled AI clusters), and integrated systems. This is HPE’s largest and most commoditized business. Revenue grew 11% in FY2025 and +33% year-over-year in Q2 FY2026, but the growth is heavily price-driven — elevated DRAM/NAND costs are passed through into average unit prices (AUPs) — while unit volumes have been supply-constrained. AI servers are sold largely to enterprise and sovereign customers (HPE deliberately avoids the lowest-margin hyperscaler volume).

Networking (~20% of revenue, ~44% of profit). The crown jewel and the strategic center of gravity. Pre-Juniper this was the Aruba campus wireless/switching franchise (built from the 2015 Aruba and 2020 Silver Peak SD-WAN deals). The July-2025 Juniper acquisition added data-center switching/routing (QFX/PTX/EX/MX), the Mist AI cloud-management platform, and service-provider relationships. This segment earns 23–25% operating margins — genuinely software-and-services-rich economics — and is the reason the stock re-rated.

Hybrid Cloud (~16% of revenue, ~9% of profit). Enterprise storage (Alletra, the former Nimble/3PAR lines), private cloud, GreenLake as-a-service software, data protection (Zerto), and infrastructure software (OpsRamp, Ezmeral). Structurally challenged: a 6% operating margin and a $1.6B goodwill impairment in FY2025.

Financial Services (~10% of revenue, ~10% of profit). A captive leasing/financing arm (HPE Financial Services) that finances customer purchases of HPE and third-party IT. It carries ~$3.5B of revenue, a stable ~10% margin, and ~$11.6B of asset-backed debt matched against financing receivables — economically a bank-like spread business bolted onto a hardware company.

Revenue model and recurrence. The bulk of HPE’s revenue is still transactional hardware (servers, switches, storage arrays) recognized at point of sale, supplemented by attached support/maintenance contracts, software subscriptions, and the GreenLake consumption/as-a-service platform, which management positions as the recurring-revenue future. Annualized recurring revenue from GreenLake and software subscriptions is growing but remains a minority of total revenue. Customers reach HPE through resellers, distributors, OEMs, ISVs, and systems integrators.

FY2026 segment realignment. Effective Q1 FY2026, HPE merged Server, Hybrid Cloud, and Financial Services into a single “Cloud & AI” segment and moved the Telco and Instant On businesses out of Networking into Corporate. Going forward the company essentially reports Cloud & AI vs. Networking — a structure that, not coincidentally, would make a future separation of the high-margin Networking business cleaner.

Verdict. HPE is a hardware-centric conglomerate whose revenue is dominated by commodity servers but whose profit is increasingly dominated by networking. It is best understood not as one business but as a high-margin networking franchise (good) stapled to a large, low-margin compute business (mediocre) and a declining storage business (bad), with a captive bank attached. The strategic logic of the last five years — and of the Juniper deal — has been to shift the profit mix toward networking. That logic is sound; the price paid and the cyclical flattery in the current numbers are the questions.

3. Industry Dynamics

HPE operates across three distinct industry structures, and they could hardly be more different in quality.

Servers — structurally bad. The x86 server market is large and, in 2025, booming: industry revenue reached record levels (IDC pegged the CY2025 server market near $444B, up ~80%, on the AI build-out). But the structure is hostile to OEMs like HPE. Roughly 47% of server units now ship ODM-direct — Quanta, Wiwynn, Foxconn and others building straight for hyperscalers, bypassing branded OEMs entirely. Among the OEMs, share is fragmented and shifting: IDC’s CY2025 reads put Dell ~10%, Supermicro ~9.5%, IEIT/Inspur ~4%, Lenovo ~4%, and HPE ~3% — i.e., HPE is the #5 branded vendor with low-single-digit global share in its single largest business. Worse, the AI portion of the market — the growth — is the least profitable: an AI server is largely a chassis wrapped around Nvidia GPUs, where Nvidia captures ~75% gross margin on the silicon and the OEM is a low-margin integrator competing on financing, supply allocation, and service. Industry AI-server gross margins run ~5%. This is a market where scale, balance-sheet, and GPU-allocation relationships matter, and where HPE is sub-scale relative to Dell and the ODMs.

Enterprise networking — structurally good (in the defended pools). Campus and data-center networking is a high-margin, oligopolistic market with real switching costs (management complexity, certified installed base, security/compliance integration). Cisco is the incumbent leader; Arista dominates high-end cloud/AI data-center switching; HPE-Juniper is now the clear #2 in campus switching (3Q25 campus revenue roughly level with Huawei’s and behind only Cisco). Margins here are 20%+ across the industry. The complication is the fastest-growing pool — AI back-end/fabric networking — where the structure is more contested: merchant silicon (Broadcom), white-box/SONiC, the Ultra Ethernet Consortium open standard, and specialists are compressing differentiation, and combined Celestica+Nvidia hold ~50% of AI back-end Ethernet. So networking is a good industry where HPE is well-positioned in the defended campus pool and weaker in the high-growth AI-fabric pool.

Storage — structurally mediocre and deflationary. Enterprise external storage is a slow-growing (~low-single-digit), deflationary market under secular pressure from public cloud and software-defined architectures. HPE is #5 at ~5.6% share and declining (~−7.5% YoY) while software-defined players take share (Pure Storage +15%, NetApp +3%). One genuine tailwind exists — Broadcom’s aggressive repricing of VMware is pushing enterprises to re-evaluate their virtualization/storage stacks, an opening HPE chases with Morpheus/VM Essentials and Alletra — but HPE’s effort is late and sub-scale, and the share trend has not reversed. The $1.6B goodwill impairment is the financial verdict on this segment.

Marathon capital-cycle lens. The server market is a textbook late-boom setup: record demand pulling in supply (ODMs, custom silicon, new capacity) with margins already cracking — the phase Marathon would avoid. Networking is a more stable oligopoly where capital discipline persists in the campus pool but is being tested in AI fabric. Storage is in slow secular decline with rational but value-less competition.

Verdict. Of HPE’s three core industries, one is genuinely attractive (networking, ~20% of revenue), one is structurally bad (servers, ~51%), and one is in managed decline (storage, ~16%). The weighted-average industry quality is mediocre, and — critically — HPE’s revenue growth is concentrated in the worst-quality pool (AI servers), which is exactly the configuration that should make an investor skeptical of headline top-line momentum.

4. Competitive Position

The honest assessment, applying the Greenwald framework segment-by-segment, is that HPE has one real moat and two commodity businesses.

Servers — no moat. Greenwald’s two diagnostic tests both fail. The market-share-stability test fails: HPE has been losing or treading water against Dell, Supermicro, and the ODMs for years; there is no incumbent stickiness producing stable share. The ROIC/margin test fails: a 7.6% segment operating margin (down from 11.4% in a single year) is the financial signature of “competitive advantages absent.” There is no proprietary technology, no switching cost, no network effect, and no cost advantage that a customer would pay up for — an enterprise can substitute a Dell PowerEdge or a Supermicro box with modest friction. HPE competes on brand, service, financing, and (currently) GPU supply allocation, none of which is a durable barrier to entry. The AI-server backlog ($16.4B cumulative orders, $5.9B backlog) is real demand, but demand is not a moat — it is a queue, and the economics of fulfilling it are thin.

Networking — a real, but localized, moat. This is where HPE earns its keep. The campus networking franchise has genuine customer-captivity and scale advantages: a large certified installed base, high switching costs (re-architecting a campus network is disruptive and risky), cloud-management lock-in via Aruba Central and Juniper’s Mist AI, and the scale to fund R&D and a global channel. The financial proof is the 23–25% operating margin sustained across years — excess returns that would erode if the moat were illusory. Juniper deepens this in data-center routing/switching and adds the Mist AIOps platform. But the moat is localized to the defended campus/enterprise pool. In the high-growth AI-fabric/data-center pool, HPE-Juniper is a merit competitor against Arista, Broadcom merchant silicon, and white-box, with no structural edge. And the DOJ antitrust settlement that cleared the deal eroded the differentiating IP: HPE had to divest the Instant On WLAN line and license Juniper’s Mist AIOps source code to a competitor — a self-inflicted dilution of the very asset that justified the premium.

Storage — no moat, share-losing. A 6% margin, a 5.6% and falling share, and a $1.6B impairment. There is no defensible position here; HPE is a price-taker losing to software-defined entrants.

Direct competitor read. Against Dell, HPE is the smaller, lower-share, lower-margin server vendor but with a better networking mix (Dell lacks an Aruba/Juniper-scale networking franchise). Against Cisco, HPE-Juniper is the clear #2 challenger in campus with a modern cloud-managed/AI-ops stack, but a fraction of Cisco’s scale and software/recurring base. Against Arista, HPE is not a serious threat in high-end cloud/AI data-center switching. Against Pure Storage/NetApp, HPE is losing share in storage. The strategic coherence of HPE’s plan is to concentrate the company around the one axis where it has a moat (networking) and to monetize/de-emphasize the rest — which is precisely what the FY2026 realignment and the Elliott-driven Strategy Committee point toward.

Verdict. HPE is a sub-scale also-ran in its two largest markets (servers, storage) with one genuinely advantaged franchise (networking, ~44% of segment profit on ~20% of revenue). Roughly 80% of revenue earns commodity economics; ~20% earns a real moat. Any thesis that pays a premium multiple for HPE must rest almost entirely on the networking franchise growing into a larger share of the whole — which is the bull case, and the reason the Juniper deal exists.

5. Growth History and Forward Opportunities

History — a decade of stagnation, then a step-change. From the 2015 spin through FY2024, HPE was a no-growth-to-low-growth company: revenue meandered from ~$28B (FY2022) to ~$30B (FY2024), and non-GAAP EPS oscillated in a ~$1.80–2.00 band for years. This is the heart of why the stock spent nine years as a single-digit-multiple value stock. Then:

(Revenue and operating income from EDGAR XBRL / 10-K; FCF from 10-K MD&A reconciliations. FY2025 operating loss reflects the $1.6B impairment.)

The FY2025 revenue jump to $34.3B (+13.8%) was Juniper (~4 months of consolidation) plus AI-server AUPs. The H1 FY2026 acceleration to +29% reflects full Juniper consolidation plus the memory-driven server ASP surge.

Quality of growth — mixed, and tilted to the lower-quality sources. Decomposing the forward inflection matters more than the headline:

• Juniper consolidation — real, durable, high-margin revenue, but acquired, not organic. It is paid for (with debt, preferred, and a drained cash balance), and the question is integration and synergy realization, not whether the revenue exists.
• AI servers — real organic demand ($16.4B cumulative orders), but low-margin, lumpy (revenue conversion keeps slipping later in the year), and concentrated in the most contested pool. Management does not disclose AI-server revenue or margin separately, which is itself a tell.
• Memory super-cycle — a cyclical price tailwind. Server revenue grew +33% in Q2 FY2026 on ASP (memory pass-through), not units. This flatters both revenue and, as memory costs are passed through with markup, margins — Q2 server margin recovered to 12.4%. This reverses when DRAM/NAND prices normalize (management itself expects elevated memory “well into 2027,” i.e., it is explicitly a multi-quarter cyclical condition, not a permanent shift).

Forward opportunities. (1) Networking mix-up: management guides Networking to >50% of operating profit and FY2027 networking margins in the mid-to-high-20s, with $200M+ of cost synergies (to be exceeded by end-FY2026) and ~$1B of combined structural savings by FY2028. (2) GreenLake / recurring revenue: the as-a-service transition, if it scales, would raise revenue quality and visibility. (3) AI inferencing at the enterprise edge: HPE’s pitch is that the next AI wave (inferencing, agentic AI, sovereign/on-prem) favors its enterprise/edge footprint over hyperscale. (4) FY2027 framework: revenue +8–12%, EPS growth +12–16%, FCF ≥$4.5B.

Verdict. The growth is real but lower-quality than the multiple implies. The durable, high-quality piece is Juniper-driven networking; the cyclical/low-margin pieces (memory windfall, AI-server pass-through) are doing a lot of the near-term work. An investor underwriting the FY2026–27 framework is implicitly betting that the high-quality piece compounds while the low-quality pieces don’t give back as much as a normalizing memory cycle would suggest.

6. Financial Quality

The GAAP/non-GAAP chasm is the first thing to understand. FY2025 GAAP was ugly: a −$437M operating loss and a net loss to common of −$59M (after $116M of preferred dividends). But this was driven almost entirely by below-segment, largely non-cash items — the $1.6B Hybrid Cloud goodwill impairment, $511M of (mostly Juniper) intangible amortization, $643M of stock-based compensation, $458M of acquisition/disposition charges, and a $275M cost-reduction program. Segment operating earnings actually rose, to $3,603M from $3,469M. On a non-GAAP basis, diluted EPS was $1.94 (vs. $1.99) — essentially flat. The operations did not deteriorate in FY2025; the GAAP result was a clean-up of prior M&A (storage) colliding with new M&A costs (Juniper).

Margins — the single most important quality signal is the server line. Server operating margin fell from 11.4% (FY2024) to 7.6% (FY2025) even as revenue grew 11% — the unmistakable signature of margin-dilutive AI-server mix and the Q1 FY2025 discounting/inventory misstep. It has since recovered to 12.4% in Q2 FY2026, but that recovery is substantially a memory-cost-pass-through effect, not proof of structural pricing power. Networking margins are the bright spot at 23.3%. Storage at 6.0% is value-less. Blended segment margin (~10.5%) is mediocre for a “technology” company and reflects the commodity weight.

Cash flow — weak and volatile. FCF fell 57% to $986M in FY2025 (OCF $2,919M less ~$2.3B of capex/capitalized software). This is a thin ~$986M of cash on a ~$60B market cap (~1.6% FCF yield on the trailing print). Management guides FY2026 FCF to ≥$3.5B and FY2027 to ≥$4.5B — a large step-up — but those figures benefit from working-capital normalization, the memory windfall, and full-year Juniper, and HPE’s FCF has historically been lumpy (it swung from $4.4B in FY2023 to $2.3B to $986M). The quality-of-earnings question is whether the guided FCF is durable run-rate or cyclically inflated; the prudent read is “partly the latter.”

Returns on capital — structurally low. With goodwill + intangibles of $30.1B now exceeding $24.7B of book equity, tangible book is deeply negative, and consolidated ROIC sits around or below the cost of capital across the cycle (the FY2025 GAAP loss makes the point starkly, but even normalized, a ~$3.6B segment profit on a ~$76B EV and a heavily-goodwilled asset base is an unexceptional return). This is not a high-return compounder; it is a capital-intensive integrator whose returns depend on the networking mix improving.

Balance sheet. Cash fell from $14.8B to $5.8B (funding the all-cash Juniper deal); total debt is ~$21.5B, but ~$11.6B is HPE Financial Services asset-backed debt matched to financing receivables (a self-liquidating spread book). Operating (ex-FS) net debt is ~$5.7B, plus the $1.5B preferred ≈ ~$7.2B of true corporate leverage — roughly 2x corporate EBITDA. Net leverage fell from 3.1x to 2.3x by Q2 FY2026, with a 2x target pulled forward to end-FY2026. So the balance sheet is stretched but improving, and deleveraging is the stated near-term capital priority — which is why the buyback is paused.

Dilution leakage. SBC jumped ~50% to $643M in FY2025, and the share count is drifting up, not down, with buybacks suspended and the preferred set to convert into ~76–93M new shares (~6%) in 2027. For a “value” stock, the share count is moving the wrong way.

Verdict. The economics do not obviously improve with scale. HPE’s quality is bifurcated: a high-quality networking core wrapped in low-quality compute/storage, with weak and volatile cash generation, structurally low returns on a goodwill-heavy balance sheet, and rising dilution. The non-GAAP P&L is far healthier than the GAAP one, but even the non-GAAP picture is of a mediocre-return business currently enjoying a cyclical tailwind.

7. Capital Allocation

Management’s capital-allocation record is the part of the thesis that should give a fundamental investor the most pause, because it is pro-cyclical and not return-disciplined.

Buybacks — bought high, paused low. HPE repurchased stock through the cycle at what now look like the wrong times: ~$213M (FY21), $512M (FY22), $421M (FY23), $150M (FY24), $202M (FY25) — meaningful repurchases when the stock was in the $13–18 range, then effectively suspended as the stock tripled to $45. The largest historical tranche (~$5.8B in FY18–19) was also at higher relative valuations. Now, at a decade-high valuation, HPE is issuing equity-linked capital ($1.5B preferred) and ~$10B of incremental debt, while diluting via SBC. This is the textbook inverse of value-accretive capital allocation: buy your own stock when it’s cheap, not when it’s dear; HPE has done the opposite.

M&A — a consistent strategy, executed at full prices. HPE has spent a decade buying networking and software to escape the server commodity trap: Aruba (2015), Nimble, SimpliVity, Silver Peak (2020), Cray ($1.3B, 2019), Zerto ($374M, 2021), OpsRamp, Athonet, and now Juniper ($13.6B total consideration, 2 July 2025). The strategic direction is sound — toward higher-margin networking — but the price discipline is questionable. Juniper was bought for ~2.6x EV/sales and ~13–14x EV/EBITDA (and >20x GAAP operating income) on ~$5.5B of revenue — a full price, above HPE’s own consolidated multiple, funded by levering up to negative tangible book. And the FY2025 $1.6B storage impairment is, in effect, a write-down of prior M&A (Nimble/3PAR-era goodwill), evidence that not all of this deal-making has created value.

Preferred and the dilution overhang. To part-fund Juniper, HPE issued $1.5B of 7.625% Series C Mandatory Convertible Preferred (30M shares, Sept 2024), which mandatorily converts ~1 Sept 2027 into ~76–93M common shares (~5.7–7.0% dilution) — and, perversely, a lower stock price maximizes the dilution. The preferred costs ~$114–116M/year in dividends, enough to have tipped FY2025 GAAP common income negative.

H3C monetization — a sensible source of cash, but a loss of earnings. HPE has fully exited its ~49% stake in the Chinese H3C joint venture (sold 30% to Unisplendour for ~$2.1B in 2024, then put the remaining ~19% for ~$1.357B, completed in H1 2026). The ~$3.5B of cumulative proceeds usefully deleverage, but HPE loses the recurring equity income (H3C contributed $245M of equity earnings in FY2023, falling to $79M in FY2025), and the FY2026 “other income & expense” outlook was cut from ~$650M to ~$420–460M as a result.

Incentive design — misaligned with shareholder value. The proxy reveals the problem. CEO Antonio Neri’s FY2025 summary-compensation-table pay was ~$23.5M, up ~10% into a GAAP-loss year. The annual incentive is 80% financial — Revenue (35%), Operating Profit (50%), ARR (15%) — but “operating profit” is the non-GAAP segment figure that excludes the $1.6B impairment and deal costs. Long-term PARSUs vest on non-GAAP net income growth, with relative TSR only a ±20% modifier. There is no ROIC, ROE, or FCF gate anywhere in the design. This is an incentive structure that rewards exactly the behavior the track record questions: adding revenue and (adjusted) operating profit through dilutive, lever-up M&A, with no penalty for the returns or cash consequences. Elliott’s pressure may improve this, but as designed it is weak.

Verdict. Management has not allocated capital intelligently by the test that matters — returns. The strategic direction (toward networking) is right and the deleveraging discipline post-Juniper is encouraging, but the execution — pro-cyclical buybacks, full-price M&A funded into negative tangible book, dilutive preferred, rising SBC, and an incentive plan with no return-on-capital gate — is the weakest leg of the thesis. The Elliott overlay (75%-of-FCF return from FY2027, possible separation) is the mechanism most likely to impose discipline.

8. Changes and Headwinds — Last Two Years

The last ~18 months have been the most eventful in HPE’s post-spin history.

• Juniper acquisition (announced Jan 2024, closed 2 July 2025). A ~$14B all-cash deal that doubled the networking franchise. The path was contested: the U.S. Department of Justice sued to block it on competition grounds (campus WLAN), and HPE settled in late June 2025 by agreeing to divest the Instant On WLAN business and license Juniper’s Mist AIOps source code — concessions that dented the deal’s strategic value. Sales forces were merged 1 Jan 2026 (integration “Phase 1” complete).
• Elliott Management activism. Elliott disclosed a >$1.5B stake (~April 2025) and pushed hard, reportedly including on the CEO. The two sides reached a cooperation agreement (16 July 2025): Robert Calderoni (KLA chairman) joined the board and chairs a new Strategy Committee conducting a value-creation/structural review (a possible Networking separation is the latent catalyst); Elliott’s Christopher Hsu joined the board in 2026; an amendment (29 May 2026) caps the board at ≤14 through the 2027 AGM. Elliott also pushed the accelerated, larger capital return.
• The FY2025 reset. A Q1 FY2025 server discounting/inventory misstep dented margins; HPE launched a cost-reduction program (~2,500 jobs, ~$350M of savings by FY2027); and it took the $1.6B storage goodwill impairment (largely Q2 FY2025).
• The forward inflection (FY2026). Three consecutive guidance raises took FY2026 non-GAAP EPS to ~$3.40 (>40% raise) and FCF to ≥$3.5B, pulling the long-term plan forward two years — driven by AI demand, the memory super-cycle, and Juniper synergies arriving ahead of schedule. Net leverage fell 3.1x→2.3x.
• Macro/operational headwinds. Tariffs (~$120M estimated impact), memory-supply constraints limiting server unit shipments (a double-edged condition — it lifts ASPs but caps volume and will reverse), and the loss of H3C equity income.
• Leadership. CEO Antonio Neri (whose seat drew Elliott scrutiny) remains; CFO Marie Myers (since Jan 2024); no announced succession plan — a governance open question given the activist pressure.

Verdict. On balance these developments strengthened the forward earnings narrative (Juniper scale, synergies, raised guidance, deleveraging) while exposing the underlying fragility (DOJ concessions, storage impairment, server misstep, activist pressure, no buyback). The thesis is genuinely better than two years ago — but a large part of why the numbers look good now is cyclical (memory) and acquisition-driven (Juniper), not evidence of a durably higher-quality core.

9. Risk Analysis (Risk Matrix)

The dominant risks are #1 (memory reversal) and #5 (de-rating) — both speak to the same core concern: today’s price and today’s earnings are flattered by a cyclical tailwind, and a normalization of either the cycle or the multiple is the main way an investor loses money here. Catastrophic-loss risk is very low; this is a large, diversified, investment-grade business — the risk is mediocrity and de-rating, not ruin.

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation here (see the labeled Claude’s Take for that). The task in this section is to characterize what the current ~$45.49 price embeds.

The two-sided multiple. HPE is simultaneously cheap against peers and dear against itself — the single most important valuation fact.

(yfinance ~2026-06-10; HPE forward P/E recomputed on the raised $3.40 FY2026 non-GAAP guide. Note that aggregator “forward P/E ~19x” figures reflect a stale $2.43 consensus that predates the guidance raises.)

On peer multiples, HPE screens as the cheapest name in the set — ~13x forward earnings and the lowest EV/EBITDA — which the bulls read as a margin-of-safety. The bears counter with three points: (1) the cheapness is deserved — HPE has the lowest operating margin (8.7%), the most leverage, the weakest FCF yield, and the lowest-quality revenue mix; (2) the $3.40 denominator is cyclically inflated by the memory windfall and full-year Juniper, so the “13x” is on peak-ish earnings; and (3) against its own 10-year history, HPE sits at the 97th percentile composite valuation (P/B 99.7th, P/S 99.7th, P/E 90.6th) — it has essentially never been this expensive relative to itself, and the sell-side mean target of $29.92 sits 34% below spot.

Embedded-expectations / reverse read. To justify ~$45 (a ~$76–78B EV), the market must underwrite roughly the following to be simultaneously true:

1. Juniper re-mixes HPE toward networking economics — Networking sustainably >50% of operating profit at mid-to-high-20s margins, with $1B of structural synergies by FY2028 landing net of preferred dividends and incremental interest.
2. Server margins do not revert to the FY2025 7.6% trough after the memory cycle normalizes — i.e., the Q2 FY2026 recovery to 12.4% proves substantially structural (mix-shift to higher-value AI/enterprise), not just a transient pass-through.
3. AI-backlog conversion accelerates ($16.4B cumulative orders convert to revenue at acceptable margins, not endlessly slipping).
4. Storage stabilizes (no further impairment; the segment carries ~0% goodwill cushion now).
5. FCF re-accelerates durably from $986M (FY2025) toward the ≥$3.5B (FY2026) and ≥$4.5B (FY2027) framework — and the capital return (75% of FCF from FY2027) materializes.

At least three of these five (memory, AI margins, storage) are substantially outside management’s control. The market is, in effect, capitalizing the good outcome on all five — which is why the stock trades at the top of its own range while the sell-side, weighting the cyclical risk more heavily, sits a third lower.

Scenario sketch (illustrative, not a target).

• Bear (memory rolls over, server margins revert to ~8–9%, AI conversion disappoints, multiple de-rates toward own-history mean): normalized non-GAAP EPS ~$2.40–2.70, a ~10–12x multiple → a materially lower equity value.
• Base (memory normalizes gradually, networking mix improves, server margins settle ~10%, EPS settles ~$2.80–3.10 mid-cycle at ~13–15x): roughly fair around the current zone.
• Bull (networking compounds >50% of profit at high-20s margins, AI inferencing scales, Strategy Committee separates Networking and it re-rates to a Cisco/Arista-type multiple, EPS sustains >$3.40): meaningful upside via mix and a sum-of-the-parts re-rating.

Sum-of-the-parts is the bull’s strongest argument. A Networking business doing ~$1.6B+ of operating profit at 23%+ margins, valued at even a discounted Cisco/Arista multiple, could be worth a large fraction of HPE’s entire current EV — implying the commodity server/storage stub is being valued at a deep discount. This is precisely the logic the Elliott Strategy Committee is built to test, and it is the most credible path to upside that doesn’t rely on the memory cycle.

Verdict. The valuation embeds successful execution on multiple fronts, several cyclical, at the top of HPE’s own historical range. It is defensible on forward peer multiples and a SOTP framing, and stretched on own-history and cash-flow-yield bases. The re-rating is largely already paid for; the remaining upside is a networking-led SOTP re-rating, and the downside is a memory-cycle/multiple normalization.

11. Variant Perception

Consensus view. The sell-side is, unusually, behind the stock — the mean target (~$30) sits well below the ~$45 price, implying analysts collectively view HPE as having overshot, even as the rating distribution (mostly Holds, a cluster of Buys) is non-committal. The de facto consensus is “good operational execution and a smart Juniper deal, but the stock has run ahead of fundamentals; the memory tailwind and AI enthusiasm are doing the lifting.” Institutional ownership is ~94%, short interest a moderate ~5.8% of float.

The strongest bull case. HPE has structurally transformed its earnings mix: networking — a genuine 23%±margin moat business — is now the majority of profit and growing, Juniper synergies are arriving ahead of schedule, AI demand provides a multi-year volume tailwind ($16.4B cumulative orders), deleveraging is ahead of plan, and an activist (Elliott) is actively driving a capital-return acceleration and a potential value-unlocking separation of Networking. On forward numbers (~13x EPS, ~6% FCF yield) it is the cheapest infrastructure name available, and a sum-of-the-parts re-rating of the networking franchise alone could be worth far more than the current price. You are buying a re-rating catalyst stack at a value multiple.

The strongest bear case. Strip the narrative and HPE is still a #5 commodity server vendor (3% share, 7.6% structural margins) and a share-losing #5 storage vendor (just impaired), with one good networking business bought at a full price and financed into negative tangible book. The current earnings are flattered by a cyclical memory windfall (server revenue +33% on price, not units) that will reverse, and by acquired (not organic) networking revenue. The stock sits at the 97th percentile of its own decade valuation; insiders have bought exactly zero shares on the open market during the entire run-up (and a director sold at ~$55); capital allocation is pro-cyclical and lacks any return-on-capital discipline; and a ~6% preferred dilution lands in 2027. When the memory cycle normalizes and FY2027’s framework (EPS +12–16%) meets a tougher comp, the multiple de-rates toward the sell-side’s $30 — and the value-trap reasserts itself.

The 3–5 assumptions that matter most.

1. Is the server-margin recovery (7.6%→12.4%) structural or a memory-cost pass-through? (Bull: structural mix-shift. Bear: transient.) — Falsifiable: watch server margins as DRAM/NAND prices normalize through FY2027.
2. Does Networking sustainably exceed 50% of operating profit at mid-to-high-20s margins post-integration? — Falsifiable: segment disclosures FY2026–27; synergy realization vs. the $1B/FY28 target.
3. Does the multiple hold at the top of its own range, or de-rate toward history/$30? — Falsifiable: the gap between price and sell-side target either closes up (targets rise) or down (price falls).
4. Does the Strategy Committee separate Networking and unlock a SOTP re-rating? — Falsifiable: board outcome over the next 12–18 months.
5. Does FCF durably re-accelerate to the framework (≥$4.5B FY2027), funding a 75%-of-FCF return? — Falsifiable: reported FCF and buyback resumption in FY2027.

Verdict. The variant-perception axis is “durable transformation vs. cyclical mirage.” The bull and bear are arguing about the same facts (Juniper is real; the memory tailwind is real; the valuation is high) and disagree on which dominates. The evidence supports a genuinely improved — but not high-quality, and now fully-priced — company.

12. Fact vs. Interpretation

13. Open Questions

1. What is AI-server revenue and gross margin, separately? HPE does not disclose it. Without it, the durability and profitability of the largest growth driver cannot be independently assessed.
2. How much of the FY2026 EPS raise is memory-cycle vs. structural? Management has not quantified the memory tailwind’s EPS contribution; this is the central swing factor for FY2027 sustainability.
3. Will the Strategy Committee recommend a Networking separation, and on what timeline? The single largest potential catalyst, entirely unresolved.
4. CEO succession. With Neri’s seat under activist scrutiny and no announced plan, who leads HPE through the Juniper integration and beyond?
5. What were the FY2025 server discounting/inventory misstep details (Q1 FY2025), and is the risk of recurrence (channel/inventory management) fully resolved? (Predates the transcripts read; would require the Q1 FY2025 10-Q.)
6. Juniper revenue synergies — beyond cost synergies ($1B structural by FY2028), is there evidence of cross-sell revenue synergy, or only “early” anecdotes?
7. GreenLake ARR scale and economics — how large and how profitable is the recurring-revenue base that underpins the “quality re-rating” argument?

14. What Must Be True (Bull and Bear, with Falsification Tests)

Bull case — what must be true. HPE’s transformation is structural, not cyclical: (a) Networking grows into >50% of operating profit at mid-to-high-20s margins as Juniper synergies (≥$1B by FY2028) land; (b) server margins hold double-digits after the memory cycle normalizes, proving an enduring mix-shift toward higher-value AI/enterprise compute; © AI backlog ($16.4B) converts to revenue at acceptable margins; (d) FCF re-accelerates to the ≥$4.5B FY2027 framework, funding a 75%-of-FCF return and buyback resumption; and (e) the Strategy Committee unlocks a sum-of-the-parts re-rating. If all hold, HPE is a genuinely re-rated, networking-led franchise that is cheap at ~13x.

Falsification test for the bull: server operating margin falls back below ~9% as DRAM/NAND prices normalize through FY2026–27, and/or Networking operating margin fails to sustain the mid-20s post-integration. Either would prove the FY2026 strength was a cyclical/accounting artifact rather than structural improvement — and the cheap-on-forward-earnings argument collapses because the forward earnings were peak.

Bear case — what must be true. The re-rating is a cyclical mirage on a mediocre core: (a) the memory super-cycle reverses and server ASPs/margins give back the FY2026 windfall; (b) AI-server economics stay structurally thin and conversion keeps slipping; © storage takes further impairment; (d) the multiple de-rates from the 97th percentile toward the sell-side’s ~$30 as the cyclical tailwind fades; and (e) capital allocation continues to leak value (dilution, full-price M&A, no return gate). If so, HPE reverts to the value trap it was for nine years.

Falsification test for the bear: HPE delivers the FY2027 framework (EPS +12–16%, FCF ≥$4.5B) with Networking sustainably >50% of operating profit at mid-to-high-20s margins and server margins holding double-digits as memory normalizes — i.e., the company grows through the memory roll-over. That would demonstrate the earnings power is durable and the bear’s “cyclical mirage” framing is wrong.

The elegance of this setup is that the same test falsifies both sides: the behavior of server margins as the memory cycle normalizes, and the durability of Networking’s profit share post-integration, will resolve the debate within roughly four to six quarters.

15. Source Appendix

The full, dated citation list appears in the Source Appendix below. Primary sources include: HPE FY2025 Form 10-K (filed 2025-12-18), FY2021–FY2024 Form 10-Ks, Q1/Q2 FY2026 Form 10-Qs, FY2025/FY2026 earnings 8-Ks, the DEF 14A proxy, and Form 4 insider filings (all via SEC EDGAR, CIK 0001645590); HPE Q4 FY2025, Q1 FY2026, and Q2 FY2026 (2026-06-01) earnings-call transcripts and conference presentations; SEC XBRL company facts; IDC and Dell’Oro server/storage/networking market-share data; and reporting on the Juniper acquisition, DOJ settlement, and Elliott Management activism. Third-party valuation/estimate aggregates (public market-data sources) are used for orientation and reconciled to filings.

The analytical body of this article carries no investment recommendation and no price target; the only position taken anywhere is in the clearly-labeled Claude’s Take block at the top, which is the author’s own subjective view. This article is general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

Hewlett Packard Enterprise Company (NYSE: HPE) · Report date 2026-06-11 Supplemental to the research memo. Fact / Interpretation / Assumption labels applied where material.

General

What thoughtful questions have other investors asked about this company? The most-asked questions cluster around four themes. (1) Memory vs. structure: how much of the FY2026 server-revenue and margin strength is a transient DRAM/NAND cost-pass-through windfall versus a durable AI/enterprise mix-shift? (2) Juniper math: will the deal’s cost synergies (~$1B structural by FY2028) and revenue synergies actually land net of the ~$10B incremental debt and $1.5B preferred, and was ~13–14x EBITDA the right price? (3) Sum-of-the-parts: is the high-margin Networking franchise (>50% of operating profit) worth more separated than embedded in a commodity-server conglomerate, and will Elliott’s Strategy Committee act on it? (4) Capital allocation/governance: why is there no ROIC/FCF gate in executive comp, why were buybacks done at $13–18 and paused at $45, and is CEO succession a risk under activist pressure?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: closer to a cyclical high on the server line. FY2026 server revenue is up +33% YoY driven by ASP (memory-cost pass-through) with units supply-constrained — a cyclical price tailwind management itself expects to persist only “into 2027.” Networking earnings, by contrast, are structurally rising (Juniper consolidation + synergies), not cyclically peaked.

Driven by external environment or internal actions? Both. External: the DRAM/NAND super-cycle and the AI capex wave. Internal: the Juniper acquisition, cost-reduction program (~$350M by FY2027), and deleveraging.

How stable are revenues? Moderately unstable. The base is transactional hardware (servers, switches, storage) recognized at point of sale — inherently cyclical with enterprise IT capex — partially offset by attached support, software subscriptions, GreenLake recurring revenue, and the Financial Services spread book. AI/sovereign deals are large and lumpy; revenue conversion timing slips quarter to quarter.

Outlook for products/services? Networking: structurally favorable (oligopoly, 23%+ margins, share #2 in campus). Servers: volume tailwind from AI but margin-thin and contested. Storage: secular decline. Net: revenue growth real, margin mix the swing factor.

How big will this market be? Server market hit record levels in 2025 (~$444B, +~80% per IDC) on AI, but ~47% is ODM-direct (un-addressable for HPE); enterprise networking is a steady oligopoly; storage is low-growth/declining. Growing (servers/networking) and shrinking (storage), heavily international (HPE sells across the Americas, EMEA, APAC, Japan).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Servers: more competitive (ODMs, custom silicon, Nvidia capturing value). Networking: stable oligopoly, but the AI-fabric pool is getting more contested (merchant silicon, white-box, UEC). Storage: persistently competitive and deflationary.

How profitable is the business (ROIC, ROE)? Fact: low and, in FY2025, negative on a GAAP basis (net loss to common). Even normalized, ~$3.6B of segment profit on a ~$76B EV and a goodwill-heavy (negative tangible book) asset base is an unexceptional return at or near the cost of capital. This is not a high-return compounder.

How profitable is the industry — competitors, barriers to entry? Networking has high barriers (installed base, switching costs, scale) and high margins. Servers have low barriers and low margins (the AI portion ~5% gross). Storage has low barriers and falling margins.

Can the business be easily understood? Yes — it is a hardware/infrastructure vendor with a captive finance arm. The complexity is in normalizing GAAP for impairments/deal costs and separating cyclical (memory) from structural earnings.

Can it be undermined by foreign low-cost labor? Partly — server manufacturing is already largely outsourced/ODM, and the commodity-server economics reflect exactly that pressure. Networking’s value is in software/IP and switching costs, which are more defensible.

Do brands matter? Modestly. “HPE/ProLiant” and “Aruba/Juniper/Mist” carry enterprise trust and channel weight, but brand is not a pricing moat in servers (Dell/Supermicro substitute readily); it matters more in networking via the certified installed base.

Nature of competition / customers’ switching costs? Servers: low switching costs, price/availability competition. Networking: high switching costs (network re-architecture risk, management-platform lock-in) — the source of the moat. Storage: moderate but eroding as software-defined/cloud lowers them.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The GreenLake recurring-revenue base and the Aruba/Juniper customer relationships have value beyond book; conversely, much of the $30.1B goodwill+intangibles is over-stated relative to economic value (hence the $1.6B storage write-down).

Off-balance-sheet liabilities? Operating leases, purchase commitments, and pension obligations (legacy HP). The HPE Financial Services book (~$11.6B asset-backed debt) is on-balance-sheet but should be analyzed separately as a matched spread business.

How conservative is the accounting? Mixed. Management leans heavily on non-GAAP measures that exclude the $1.6B impairment, $643M SBC, $511M amortization, and deal costs — and ties incentives to those non-GAAP figures. The impairment itself was appropriately taken. Interpretation: read GAAP and non-GAAP side by side; neither alone is the truth.

How CapEx-hungry is the business? Moderately — FY2025 capex/capitalized software was ~$2.3B (~7% of revenue), elevated by AI-infrastructure and GreenLake assets. The Financial Services book also absorbs capital (financing receivables).

Capital Allocation & Management

How much FCF, and how is it used? FY2025 FCF was a weak $986M (down 57%); guided to ≥$3.5B FY2026 and ≥$4.5B FY2027. Recent use: funding Juniper (cash + debt + preferred) and deleveraging. Forward plan (Elliott-influenced): return ~75% of FCF from FY2027 via dividends and resumed buybacks once net leverage hits 2x (targeted end-FY2026).

Significant acquisitions recently? Yes — Juniper Networks, ~$13.6B, closed 2 July 2025 (~2.6x sales, ~13–14x EBITDA). Prior: Aruba, Cray, Silver Peak, Zerto, OpsRamp, Athonet. Interpretation: right direction (toward networking), full prices, mixed returns (FY2025 storage impairment writes down older M&A).

Buying back shares? Not currently — buybacks were ~$150–512M/year FY2021–25 and are now suspended to fund Juniper/deleveraging. Interpretation: pro-cyclical (bought at $13–18, paused at $45).

Issuing large amounts of stock to insiders? SBC rose ~50% to $643M in FY2025, and the $1.5B mandatory-convertible preferred will issue ~76–93M shares (~6% dilution) in 2027. Share count is drifting up.

Compensation policy / incentive alignment? Fact: CEO Neri FY2025 SCT pay ~$23.5M. Annual incentive 80% financial (Revenue 35% / non-GAAP Operating Profit 50% / ARR 15%); PARSUs on non-GAAP net-income growth with a ±20% relative-TSR modifier. No ROIC/ROE/FCF gate. Interpretation: weak alignment — rewards revenue and adjusted profit growth via dilutive M&A without a return-on-capital constraint.

Motivations of management? Escape the commodity-server trap by re-mixing toward networking and as-a-service recurring revenue — strategically rational. Under Elliott pressure, also focused on deleveraging, capital return, and a possible structural review.

Valuation & Market Data

ADR / MLP / K-1? No — HPE is a U.S.-domiciled C-corp common stock on the NYSE; standard 1099 treatment.

Dividend policy? $0.13/quarter (~$0.52 annualized, ~1.2% yield, ~25% payout); management aspires to be a “dividend-growth company” and to layer buybacks on top from FY2027.

How profitable is the business? Operating margin ~8–10% blended (Networking 23% / Server ~8–12% / Storage 6%); ROIC near cost of capital; FY2025 GAAP net loss to common.

Net income diverging from cash from operations? Yes, materially in FY2025: GAAP net loss to common −$59M vs. OCF $2,919M — the gap is the non-cash $1.6B impairment, $643M SBC, and $511M amortization. Interpretation: OCF is the better read of FY2025; but FCF (after ~$2.3B capex) was still a thin $986M.

Risks & Downside

What factors would cause the stock to decline? (1) Memory cycle reverses and server margins give back the windfall; (2) multiple de-rates from the 97th-percentile own-valuation toward the ~$30 sell-side mean; (3) Juniper synergies/AI conversion disappoint; (4) further storage impairment; (5) preferred dilution; (6) enterprise IT capex downturn.

Risk of catastrophic loss? Low. Diversified $34B-revenue, investment-grade franchise with a captive finance arm.

Chance of a total loss? Negligible. No plausible zero — the risk here is de-rating and mediocrity, not insolvency.

Recent News & Events

Has the business environment changed recently? Yes, materially: the Juniper acquisition closed (2 July 2025) after a DOJ settlement (divest Instant On, license Mist source); Elliott Management took a >$1.5B stake and reached a cooperation agreement (16 July 2025) creating a Strategy Committee; FY2026 guidance was raised >40% on AI + memory + synergies; H3C was fully divested (~$3.5B cumulative proceeds); net leverage fell 3.1x→2.3x.

Significant acquisitions / divestitures? Acquired Juniper (~$13.6B); divested H3C stake (China JV) and Instant On WLAN (DOJ remedy).

Change in accounting policies? No material change; FY2026 brought a segment realignment (Server+Hybrid Cloud+FS merged into “Cloud & AI”; Telco/Instant On moved to Corporate) — a reporting change that also eases a potential future Networking separation.

Recent changes — new markets, facilities, management? CFO Marie Myers (since Jan 2024); new board members (Calderoni as chair of the Strategy Committee; Elliott’s Hsu); ~2,500-job cost-reduction program; deepened AI-server and AI-fabric networking push.

APPENDIX B — Source Appendix

Hewlett Packard Enterprise Company (NYSE: HPE) · Report date 2026-06-11

Primary sources first. All financial figures reconciled to SEC filings (EDGAR, CIK 0001645590) and HPE earnings materials; third-party aggregates used for orientation only and reconciled to filings. Accessed June 2026.

Primary — SEC filings (EDGAR, CIK 0001645590)

Primary — Earnings calls & investor materials

Primary/secondary — corporate events

Industry & market data

Third-party aggregates (orientation only, reconciled to filings)

Where third-party scores or estimates are cited, the underlying primary source governs.