NTAPNetApp, Inc.

NetApp (NTAP) — research dossier, 2026-09-20

Drafted by Claude in a supervised session, no owner input. Nothing here is human-verified. Retrieved sources are named per item. Items marked [background] come from Claude's prior knowledge, not from a document retrieved this session. Items marked [web summary] come from a search-engine summary of a page that was not itself fetched.

Access limits this session

• FMP connector: only profile-symbol and peers allowed on the current plan; statements, quote, analyst, insiders, 13F, news and filings denied. Financial statements came from the FMP feed inside the Bigdata.com company tearsheet instead.
• Alpha Vantage: EARNINGS_CALL_TRANSCRIPT for 2027Q1 returned an empty transcript; the one permitted step-back (2026Q4) returned the full Q4 FY26 call. The Q1 FY27 call was read in full from Quartr via Bigdata.com. Insider transactions and the company overview were also pulled from Alpha Vantage (25-call daily cap; one rate-limit hit).
• sec.gov was reachable: the 2026 proxy, two 8-Ks and the EDGAR comment-letter index were fetched. Blocked by the egress policy: ortex.news, nasdaq.com, stocktitan.net, blocksandfiles.com, investing.com.
• Bigdata.com returned no broker research (INVESTMENT-RESEARCH) for NetApp; sell-side views below come from The Fly / MT Newswires summaries of notes.
• NetApp was not in ledger/subjects.yaml (added today) and no podcast row on the board mentions it. The board contributes nothing yet.

Snapshot (price data as of 2026-09-18 close, FMP via Bigdata.com)

Five-year financials (GAAP, FMP via Bigdata.com; fiscal years end late April)

Share count fell 13% in four years; NetApp is the opposite of a diluter. SBC is 5.5% of revenue and roughly a fifth of FCF. FY26 capex $198m (10-K). Equity is only $1.35bn against $2.9bn goodwill because of two decades of buybacks; that is a capital-return artefact, not a solvency issue.

FY26 and Q1 FY27 results (non-GAAP unless stated; Q4 call 2026-05-28, Q1 call 2026-09-02, 10-K, 10-Q)

FY27 guidance (raised 2026-09-02): revenue $7.975–8.225bn (midpoint $8.1bn, +17%, up $650m from the May guide of $7.45bn); gross margin 68.1–69.1%; operating margin 30.3–31.3%; EPS $9.73–10.03 (+22%); GAAP EPS $7.35–7.65; return up to 100% of FCF; share count down low single digits. Q2 guide: revenue $2.1bn ± $75m (+23%), gross margin 67–68%, operating margin 30.9–31.9%, EPS $2.54–2.64. The CFO said 1H/2H is "roughly 50/50" ex the extra week, which by Evercore's arithmetic on the call implies second-half growth of about 9–10%. Long-term targets restated: product GM mid-to-high 50s, public cloud GM 80–85%.

The stock fell about 9% after hours on the print and recovered within days. The pushback, in the sell-side notes Bigdata.com carried: pull-forward (Morgan Stanley, "strong enough to raise estimates, but not the multiple"), a sequential gross-margin dip and decelerating guide (Wedbush, Citi cut PT to $190), component costs (BofA), and no "clear transformational aspect" to the model (Wedbush). Northland: "customers absorbing cost raises were a big factor... but data points suggest NetApp is still a share gainer."

Management's own words on the two contested points (Q1 call, Q&A):

• Pull-forward: "certain transactions that we expected to be built out over multiple quarters happening within a quarter... 'Let's do two of the four [data centers] that we want to do faster this calendar year'... It is not a material part of the overall business."
• Pricing: "in the past... it would take probably two to three quarters to start seeing it, but now we're seeing it much earlier." Customers "budget in dollars"; some moved lower-value workloads from all-flash to hybrid flash.

Pipeline scorecard

Lives in scorecard.md next to this file and is updated there; this dossier is a dated snapshot.

S0 detail — ❌ FAILED, 1/4

1. Distribution lock-in = 1. ONTAP installed base with a $2.6bn support annuity at 93% gross margin; $4.85bn deferred revenue; first-party services inside AWS (FSx for ONTAP), Azure (Azure NetApp Files) and Google Cloud; "the only enterprise storage platform natively embedded in the world's largest clouds" (company release, a claim). The CEO's "a significant share of the world's enterprise unstructured data resides on NetApp solutions" (Q4 call) is also a claim, unquantified.
2. Unscrapeable data = 0. The data on NetApp arrays belongs to customers. NetApp's own fleet telemetry (AutoSupport / Active IQ) is [background] and no document this session shows it conferring an advantage. If the owner accepts it, the score is 2/4, still a fail.
3. Liability absorption = 0. NetApp sells systems and software; Keystone sells capacity with SLAs but sits inside a $407m professional-services line.
4. Bottleneck tokens can't manufacture = 0. NetApp does not own the scarce physical input; it buys NAND and DRAM from "a limited number of suppliers" and builds through "a few key contract manufacturers" (10-K). ONTAP is thirty years of engineering, which is a moat against start-ups but not a physical bottleneck.

Surplus question: today NetApp is keeping some of the surplus. Product GM held 54.6% against an implied low-50s guide because price increases landed within a quarter instead of two to three. But that is cyclical pricing power in a shortage, shared with Everpure (which raised list prices about 20% in February 2026 [web summary]), Dell and HPE. When NAND supply normalises, the customer keeps the savings; NetApp's structural pricing power sits only in the support renewal. Falsifier: NAND and DRAM capacity arrives (HPE: wafer capacity is the structural fix, "should improve some in 2027"), ASPs deflate, and NetApp is holding $1.9bn of non-cancelable purchase commitments at prices "established in advance of delivery" (10-Q risk factor). Liquidity: $600m a day, no constraint. Competence: NetApp sells the ONTAP storage operating system and the flash and disk arrays it runs on, to enterprises directly and through two big distributors, and rents the same software as a service inside the three big clouds; it earns a support annuity on the installed base. Exclusion list: ⏳ OPEN (user).

S2 — kill test passes

Revenue reaccelerating from 5% to a 17% guide; gross margin ~70%; net cash; FCF $1.87bn (27% of revenue) in FY26; ROIC about 19–21% (FMP key metrics); share count shrinking 3% a year; 14–71 analysts depending on the feed, so no coverage edge. Flags to carry: Q1 FCF -35% y/y on inventory; total purchase commitments $2.3bn ($2.1bn due within twelve months) against $1.4bn at FY-end; two distributors are 43% of revenue.

S3: the three biggest risks management itself discloses (10-K FY26, 10-Q Q1 FY27)

1. Customer and channel concentration: two distributors ("Customer A" 22%, "Customer B" 21%) with reseller agreements that "generally do not require minimum purchases"; most sales are fulfilled indirectly, which "makes it particularly difficult to predict future revenue."
2. Supply: reliance on a limited number of suppliers and a few contract manufacturers; firm, non-cancelable commitments at pre-set prices, so "if market prices for such materials or components decline, we could be obligated to acquire inventory at costs that exceed prevailing market prices."
3. Competition from "full-stack vendors," "newer public companies focused on flash storage," "new market entrants targeting the AI opportunity," and cloud providers that "are also our competitors."

Also disclosed: cybersecurity risk factor explicitly naming AI-assisted attackers ("Increasing use of AI, such as Anthropic's Claude Mythos, in techniques employed by threat actors"); a 14-week Q1 that flatters year-on-year comparisons; $2.3bn of the $3.6bn cash held abroad. Red flags checked: auditor Deloitte, no change; one CAM (evaluation of performance obligations); no going-concern or restatement language seen; no NT 10-K/10-Q on EDGAR; last SEC staff comment letter 2023-04-10 with the company's last response 2023-03-10 (EDGAR index, fetched); ongoing non-practising-entity patent suits with no accrual; the 2019 securities class action is historical [web summary]. Proxy: FY26 CEO pay $22.16m with 95% at risk [web summary]; annual bonus 50% on operating income after SBC, 30% revenue, 10% cloud storage revenue, 10% block-storage growth; PBRSUs move from Billings to Revenue plus non-GAAP EPS for FY27, which pays management for exactly the pass-through-driven revenue this cycle produces. Say-on-pay 93.9% in 2025 and about 93% in 2026. Officer-exculpation charter amendment passed 2026-09-09 with 21.6m against. Director Nevens drew 19.1m votes against, the most of any nominee; the proxy was not read for why.

S4: value chain in words

NAND, DRAM, CPUs, HDDs from unnamed suppliers (industry: Samsung, SK hynix, Kioxia/SanDisk, Micron [background]) → contract manufacturers in Fremont, San Jose, Laredo, Guadalajara, Helmond, Tiszaújváros, Taoyuan and Singapore (10-K [web summary]) → NetApp (ONTAP; AFF A/C-series, AFX disaggregated, FAS hybrid, ASA block, StorageGRID object; Keystone; first-party cloud services) → two distributors at 43% plus VARs, SIs, OEMs and hyperscaler marketplaces → enterprises, public sector, neoclouds, sovereign clouds. Pricing power sits at the pass-through point during a shortage and, structurally, at the support renewal. Market position: IDC Q1 CY2026 external storage $9.2bn +22.7% with all-flash above 50% of revenue for the first time; ranking Dell, NetApp, Everpure, Huawei, HPE; for CY2025 Dell 23.7%, Huawei 12.6%, NetApp 8.1%, Everpure 7.1% [web summary of blocksandfiles.com]. Gartner 2026 Magic Quadrant (2026-08-19): six Leaders, Everpure highest on both axes, NetApp and Dell "mid-quadrant" [web summary]. Cross-check that cannot all be true: NetApp says it is displacing "legacy and flash-only competitors," Everpure claims "outsized market share gains," Dell says its IP storage has grown "ahead of the market for six consecutive quarters." Some of the growth every vendor is reporting is price.

S5: who is on the other side

Passive core: BlackRock 9.9%, two Vanguard entities 13.8% combined, State Street 5.2% (proxy, as of 2026-07-13). Q2 CY26 13F changes (WSRA via Bigdata.com): holders up 100 to 965 with 183 new positions; Norges Bank new at 4.2m shares; quant and multi-strategy buyers (Jane Street, Citadel, Two Sigma, Arrowstreet, Lazard); sellers are value managers (Beutel Goodman -47%, Ameriprise -22%, Invesco -21%, Wellington -17%, Victory -31%). That is a value-to-momentum handoff at a doubled share price. Insiders own under 1% and have sold into strength all year with no open-market buys: the CEO sold about 40,000 shares (~$7.7m) on 2026-09-14 at $190–194, the President ~49,000 (~$7.6m) on 2026-06-23, two directors in June and September; Alpha Vantage does not say which were 10b5-1 sales. Shorts covered from 14.6m to 9.4m shares into the print [web summary]. Capital structure: automatic shelf on file but the company is a net buyer of its stock; $550m of 2.375% notes fall due June 2027 and will likely be refinanced at roughly double the coupon (the 2032/2035 notes cost 5.5–5.7%).

S6: outside-in evidence

• Gartner Peer Insights: NetApp 4.8/5 on 934 reviews vs Everpure 4.9 on 714; dislikes include limited price-negotiation windows and support [web summary].
• Glassdoor: 3.8/5 on 3,639 reviews, 67% recommend, 67% CEO approval; work-life balance down 6% in a year; repeated complaints about twice-yearly layoffs and a forced ~10% "not meeting expectations" quota [web summary]. Comparably CEO score 78/100 [web summary].
• Job postings (LinkUp via Bigdata.com): 381 average active in the trailing twelve months, +20% y/y; August 2026 416, +14% y/y; the last thirty days 369, -10% versus the prior thirty.
• Peer calls (Quartr via Bigdata.com): Everpure's CEO, "In my over 40 years in technology, I have never seen another supply chain situation remotely like this"; customers "paying more for less capacity"; Dell and HPE both report record or near-record storage on the same AI-modernisation story with DRAM/NAND/CPU/HDD constraints. NetApp's quarter is partly industry beta.
• VAST Data raised $1bn at a $30bn valuation in March 2026 and sells to neoclouds and xAI [web summary]; a Sacra summary says it displaces NetApp and Pure at refresh; both VAST and NetApp claim NVIDIA SuperPOD certification, so treat both as marketing.
• Layoffs: 77 San Jose roles cut 2026-05-26 per WARN summaries [web summary] against 11,700 employees (10-K).
• Expert or customer call: ⏳ OPEN (user).

S7: draft thesis (not the owner's, not scored)

Baker lens, in his terms: this is a true capacity cycle in memory, not an inventory cycle. The sellers of shortage are the memory makers; NetApp is a buyer of shortage that is, for now, passing the cost through faster than in past cycles and enjoying a demand response that runs opposite to every previous price spike ("we saw the opposite," CEO). Installed-base logic cuts the wrong way for a systems vendor: in a shortage customers can sweat what they have, and the fact that they are not is the whole bull case. Different-and-hard: unified file/block/object under one OS with the same software native in three hyperscalers is different and took thirty years; AFX disaggregation is the answer to VAST's architecture, one generation in. Token path: NetApp is adjacent to it (AI Data Engine, DataPelago) but not in it.

Bull: consensus is "Hold" at 20x a guide management has raised by $650m one quarter in; the durable pieces (first-party cloud storage +30% in FY26, Keystone +65%, unbilled RPO +88%, support at 93% GM) are compounding underneath a cyclical print; IDC has NetApp back at #2; opex discipline turns any revenue upside into 30%-plus margins and 100% of FCF comes back as buybacks at a 5% yield.

Bear: the 30% quarter is a cocktail of a 14th week, industry-wide price inflation, some pull-forward and genuine AI modernisation, in unknown proportions; the second-half guide of roughly 9–10% is the run rate; when wafer capacity lands in 2027 ASPs deflate, the $1.9bn of committed inventory is marked against falling prices, product GM revisits the low 50s, and revenue growth turns negative on price alone while the stock still trades on peak earnings. Baker: a 60–70% price spike with record margins "is the time to sell — except in a true capacity cycle," and he would ask which one this is before paying 20x.

Kill criteria (also in scorecard.md): product GM below 50% for two consecutive quarters; AFA growth below 10% while Everpure and Dell IP storage grow faster; public cloud below 12% ex-week; inventory and commitments rising into a guide cut; the two distributors above 50% or a distributor change.

S8: valuation frame (owner to decide)

EV about $37.7bn ($38.8bn market cap less $1.1bn net cash). On the FY27 guide midpoint: 20.0x EPS of $9.88, 4.7x EV/revenue, about 14.4x EV/EBITDA on the $2.62bn consensus. FY28 consensus EPS $11.10 (12 analysts) → 17.8x; FY29 $12.30 on only two analysts. FCF yield on FY26 4.8%. Scenario sketch for April 2029, illustrative only: bear EPS $9.00 (price normalisation, 3–5% growth, 66% GM) at 14x = $126, about -36%; base $12.30 at 17x = $209, roughly flat plus 1% dividends; bull $14.50 at 20x = $290, about +47%. That is not a fat pitch at $198 after a double in six months.

Druckenmiller lens: the puck 18–24 months out is a memory capacity add, and the trade is swimming with a shortage that everyone on every call already sees. Death-knells check not run this session; the one data point retrieved is Brent above $97 on 2026-09-03 (Ifeng market wrap via Bigdata.com), so oil is rising and the dollar and rates were not checked. Proposal, for the owner to accept or reject: no chase at the 52-week high; a starter of at most 1.5% of the portfolio only if the owner holds the share-gain-plus-cloud-mix variant view, entered under about $185 (the 50-day average) with a second tranche only after the 2026-12-01 call shows product GM at or above the 54–55% Q1 level and purchase commitments flat; hard cap 3%. If the owner does not hold that variant view, the honest answer is a watchlist entry.

S9: monitoring

INSIGHT 2026-09-29 (cloud AI launches, hyperscaler block storage price points); Q2 FY27 call 2026-12-01 (product GM against the 67–68% total-GM guide, inventory, commitments, the "50/50" halves claim); Everpure's Q3 FY27 call in late November for the share comparison; TrendForce NAND and DRAM contract-price prints each quarter as the leading indicator (DRAM contract +90–95% q/q in Q1 CY26 [web summary]); LinkUp job postings monthly; 13F for Q3 CY26 (Norges, the value sellers); Form 4s for the CEO's selling cadence; the June 2027 note refinancing; IDC's Q2 and Q3 CY26 trackers; the annual proxy for whether the PBRSU metrics shift again.

Connection map seed

See relationships.yaml: 55 nodes, 66 sourced edges (officers, all ten directors with vote counts, 5% holders and 13F movers, insider sales, four note series and the buyback, auditor and CAM, two acquisitions and one divestiture, two unnamed distributors at 43%, supplier commitments, hyperscaler and NVIDIA partnerships, six competitors, SEC comment-letter history, litigation watch).

This is process support, not investment advice.