NetEase, Inc. (NASDAQ: NTES) — Winning the Android Tax Revolt, Hoarding the Spoils

An independent, evidence-first equity research note. Report date: 2026-08-01. Price reference: US$133.27 per ADS, close 2026-07-31. Each ADS represents five ordinary shares. NetEase reports in RMB; unless a US$ figure is the company’s own convenience translation, conversions use RMB6.8980 = US$1.00 (the rate in the Q1-2026 release) or RMB6.9931 (the FY2025 20-F rate), and every trend is computed in RMB.

The analysis below takes no position and contains no price target anywhere. The single, deliberate exception is the Claude's Take block immediately below.

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information and commentary, not investment advice, and it is not a recommendation to buy or sell any security. Everything below it — the Business Overview through What Must Be True sections — is deliberately position-free and carries no price target.

Verdict: BUY — accumulate here, add hard below ~$120. Conviction: medium-high. My fair-value zone is ~$160–185 per ADS (roughly 17–19x cash-adjusted non-GAAP earnings plus the cash); the accumulation zone is ~$115–140, which is where the stock sits today; above ~$200 I would be trimming rather than buying. This is the cheapest large, genuinely-moated content business I have looked at this year, and the reason it is cheap is a jurisdiction, not a business.

The arithmetic that decides it: at $133.27, NetEase carries US$24.3bn of company-defined net cash — 28.5% of the market capitalisation, $38.03 per ADS. Strip it out and you are paying $95.24 for $7.70 of non-GAAP operating earnings per ADS — 12.4x — for the world’s second-largest games developer, running a 33% operating margin, on negative operating invested capital, with a 45.5%-owner founder who has never sold a share. A no-growth perpetuity on current after-tax operating profit at a 9% cost of equity is worth 84% of today’s enterprise value: the market is underwriting about 1.5% of perpetual real growth from a company that grew operating profit 21% year-on-year last quarter. And the reason profit is growing is the most interesting thing in the file. Since late 2024 NetEase has been progressively pulling its biggest franchises — Fantasy Westward Journey, Onmyoji, Sword of Justice, Minecraft — off the Huawei, OPPO, vivo and Xiaomi Android stores that have historically taken up to 50% of every yuan spent, forcing players onto its own distribution. In Q1-2026 the games-segment gross margin jumped 68.8% → 74.8%, cost of revenues fell 9.4% in absolute terms on 6.1% higher revenue, and operating margin hit 41.4%. On top of that, Apple cut its China App Store commission from 30% to 25% on March 15, 2026 — two weeks before quarter-end, so it is barely in the numbers yet. This is a content oligopolist expropriating the rents of its own distribution layer while the antitrust regulator squeezes the one foreign gatekeeper. It is a supply-side, capital-cycle story, and it is structural.

The framing is contrarian value with an identified catalyst, and the tape agrees rather than argues. NetEase carries a negative Momentum loading (−0.14) into a regime where Value is +16.1% over twelve months (z +1.88) and the China-internet basket it loads on hardest (+1.31, its single largest factor) is −13.1% over the same window and only now turning. Sixty-nine percent of its variance is idiosyncratic. Every one of its ten nearest factor-neighbours is a Chinese ADR and not one is a games company — the model classifies NetEase by the exchange it lists on, not the business it runs, which is the mispricing stated quantitatively. The catalyst is dated: the dual-primary Hong Kong listing completed 2026-06-25, and the stock rose 7.7% and 4.4% in the two sessions that followed.

What keeps this at medium-high rather than high conviction is capital allocation and jurisdiction. NetEase produced US$6.96bn of free cash flow in 2025, returned US$2.07bn, and cut buybacks 92% to US$91m while the share count rose 0.76% — it extended a US$5bn authorisation it will not use, and let the cash pile grow instead. That is not misallocation; it is non-allocation, and it is the difference between this being cheap and this being a compounder. Meanwhile 85.2% of revenue is generated inside variable-interest entities that shareholders do not own, the SEC had to force management to restore its own “the PRC government may intervene at any time” risk language in 2025, and this stock has twice fallen more than 11% in a single session on a Beijing announcement. Catchy version: you are buying China’s second-best content franchise at a Chinese-ADR multiple, with a quarter of the price refunded in cash — and the founder holding 45% alongside you.

Conviction: medium-high. The single piece of evidence that would make me more bullish: the US$5bn buyback actually being used — even US$1.5bn a year would signal that management shares the view that its own stock is cheap, and would convert this from cheap-asset to compounder. The single piece that would flip me bearish: games gross margin falling back below 70% for two consecutive quarters while revenue decelerates — that would say the Android-tax victory was a one-quarter timing artifact and the market was right to pay 67th-percentile price-to-sales for it. A second, faster kill switch: any revival of the December-2023 draft rules on in-game spending caps.

📈 Stock Price Action — Five-Year Event Map

Over five years NetEase has round-tripped and then some: from $106 in mid-2021 to a low of $50.57 on 2022-10-28, to an all-time high of $156.20 on 2025-09-17, to $133.27 today — 14.7% below that high, 163.5% above that low, and inside a 52-week range of $108.62–$156.20. The 21-, 50- and 200-day EMAs sit at $127.07 / $124.69 / $122.91, all rising, with price above all three. Compounded, the five years have delivered 7.3% a year for 44% volatility and a −51.4% maximum drawdown — a 0.12 Sharpe. The ten-year figure is 14.9% a year. The five-year window is, almost entirely, the story of what Beijing did to this sector in 2021–22 and how much of it has since been given back.

Cycle narrative. (1) The 2021 leg was regulatory, not operational: revenue never fell, but the market repriced the right to earn it after state media called online games “spiritual opium” and the NPPA restricted minors to three hours of play a week. (2) The 2022 trough compounded that with the delisting threat under the Holding Foreign Companies Accountable Act; at $50.57 NetEase traded at roughly book value plus its cash. (3) The +25.7% session of 2022-03-16 is worth remembering as a risk statement in both directions — this security can move a quarter of its value in a day on a policy sentence. (4–6) The recovery was earned: approvals resumed, the Blizzard catalogue came back under Microsoft in 2024, and for the first time NetEase produced genuine global hits rather than exports — Marvel Rivals at 40 million registered players inside two months and Where Winds Meet from 15 million players to more than 80 million within a year. (7) The drawdown from the September 2025 high was a GAAP-quality de-rating, not an operating one: operating profit held while investment marks and currency losses cut reported net income by a third. (8) The current leg has two dated causes — the Q1-2026 margin step-change on 2026-05-21, and the completion of the dual-primary Hong Kong listing on 2026-06-25, which opens the door to Stock Connect southbound flows and removes the ADR-delisting tail. Price moves are facts; the attributions above are interpretation, cross-referenced to 6-K filing dates where one exists.

1. Executive Summary

NetEase is China’s second-largest games company and, by revenue, one of the three largest in the world. In FY2025 it earned RMB112.6bn (US$16.1bn) of revenue, RMB72.4bn of gross profit (64.3%), and RMB33.8bn of net income attributable to shareholders. Games and related value-added services are 81.9% of revenue and 89% of gross profit; the three non-games segments — Youdao (intelligent learning, NYSE: DAO), NetEase Cloud Music (HKEX: 9899) and an innovative-businesses bucket containing Yanxuan — together contribute 18.1% of revenue and are, in aggregate, roughly break-even and shrinking.

Three facts define the current investment situation.

First, the margin structure changed in the last two quarters, and the cause is identifiable. In Q1-2026 cost of revenues fell 9.4% year-on-year in absolute RMB on revenue that rose 6.1%. Games-segment gross margin went from 68.8% to 74.8%; total gross margin from 64.1% to 69.4%; operating margin from 36.2% to 41.4%. The company’s stated cause is “lower revenue sharing costs related to the platform,” and the mechanism is public: since late 2024 NetEase has withdrawn its largest franchises from the Chinese Android app stores that demand up to 50% of in-app spending, forcing users onto its own distribution. Separately, Apple cut its China App Store commission from 30% to 25% on 2026-03-15, which is present in only two weeks of the quarter.

Second, the balance sheet is extraordinary and under-used. Net cash at 2026-03-31 was RMB167.5bn (US$24.3bn) — 28.5% of market capitalisation, $38.03 per ADS. Operating invested capital is negative: RMB21.8bn of customer prepayments (contract liabilities) plus payables exceed the operating asset base, so the games business consumes no net capital. Capital expenditure is 1.8% of revenue. Against that, FY2025 buybacks collapsed 92% to US$91m, the share count rose 0.76%, and management extended to 2029 a US$5.0bn authorisation of which only US$2.1bn has been used in 3.2 years.

Third, everything is inside one jurisdiction and one legal structure. 85.2% of revenue is generated by variable-interest entities in which NetEase holds no equity, consolidated on contract. The SEC’s Office of Technology required NetEase in 2025 to restore risk language it had softened, forcing it to state that “the PRC government may also intervene with or influence our operations as it deems appropriate to further regulatory, political and societal goals at any time.” Twice in five years a Beijing announcement has taken more than 11% off the stock in a single session.

Valuation reflects the third fact far more than the first two. On hand-built figures — vendor enterprise-value and price-to-book data for this ticker are wrong by a factor of five, for reasons documented in the valuation section — NetEase trades at 17.4x trailing GAAP earnings, 15.9x non-GAAP, 11.0x EV/EBIT, 3.67x EV/sales, and 13.6x cash-adjusted GAAP earnings (12.4x non-GAAP). That is in line with Tencent, at roughly half the multiple of Nintendo and Square Enix, and it implies about 1.5% of perpetual real growth in operating profit. On its own ten-year history the stock sits in the 21st percentile on P/E and the 25th on P/B — but the 67th on P/S, because the net margin has doubled from 16.4% (FY2020) to 34.9% (Q1-2026). Which of those percentiles is the true reading is the same question as whether the Android-tax victory holds. This note argues the evidence favours durability, and quantifies what happens if it does not.

2. Business Overview

2.1 What NetEase actually sells

NetEase sells recurring in-game monetisation on a portfolio of more than one hundred live-service titles, the majority of which it wrote itself, the oldest of which have been generating revenue for more than twenty years. Everything else the company does is either an adjacency built off that cash flow or a legacy of its origins as a 1997 web portal.

The revenue model is free-to-play with in-game item, cosmetic and battle-pass monetisation, plus subscriptions, plus a smaller premium/console component. Because players top up wallets and buy passes in advance of consuming the content, revenue is deferred: contract liabilities stood at RMB21.8bn (US$3.2bn) at 2026-03-31, up from RMB20.5bn three months earlier. That float is the mechanical reason operating cash flow runs at 1.3–1.5x net income and the reason invested capital is negative.

2.2 Segments and their economics

The table understates how concentrated the profit is. Games generated RMB64.2bn of the RMB72.4bn of FY2025 gross profit — 88.7%. The other three segments together produced RMB8.2bn of gross profit against a corporate operating-expense base of RMB36.6bn; on any sensible allocation they are collectively at or below break-even. NetEase is a games company with three appendages, two of which are separately listed and one of which (innovative businesses) is being deliberately shrunk — its revenue fell 15.9% in 2025, partly through a RMB961.1m increase in inter-segment eliminations and partly through declining product sales at the Yanxuan private-label e-commerce operation.

Within games, the split that matters is self-developed versus licensed. In-house developed games generated RMB78.9bn in FY2025, +8.6% — 86% of the segment and 70% of total company revenue. Licensed titles (principally the Blizzard catalogue and Mojang’s Minecraft) were 9.5% of total revenue in 2025, up from 7.4% in 2024 and 4.6% in 2023 as the Blizzard portfolio returned. That growth is welcome for the top line and unhelpful for the margin: FY2025 games cost of revenues rose 6.8% “driven by higher royalty fees as a result of increased revenue from licensed games,” and the 20-F states plainly that licensed arrangements “provide us with relatively less profit than games that we develop in-house.”

2.3 The portfolio

NetEase’s franchise list is unusual in the industry for its age. The Fantasy Westward Journey family — PC and mobile — has been operating for more than two decades and in Q1-2026 set a record 3.9 million peak concurrent users on the PC “unlimited server,” a figure management attributes to a deliberate re-architecting of the progression system to bring back lapsed players. On the same call management cited an MMO franchise “with over 18 years of operations” delivering record quarterly commercial performance, a 2016-launched MMORPG passing 200 million cumulative registered users at its tenth anniversary, and an eleven-year-old title reaching its highest DAU level since 2017.

Alongside those sit the newer domestic hits — Identity V, Eggy Party (China’s leading party game), Sword of Justice, Infinite Borders, Naraka: Bladepoint — and, critically, the first genuinely global NetEase products: Marvel Rivals, which reached 40 million registered players within two months of its December 2024 console/PC launch, and Where Winds Meet, a cross-platform (PC, mobile, console) open-world title that went from 15 million players in Q4-2024 to more than 80 million cumulative players by end-FY2025, holds a 78% positive Steam rating, and reached #2 on Steam’s global top-seller chart on its version 1.6 update in Q1-2026.

The pipeline named on the Q1-2026 call is two titles: Sea of Elements, in final preparation with a targeted Q3-2026 launch window, and Ananta, an urban open-world game still being refined with no announced date. Management’s framing of Ananta — “we do not shy away from competition, but our focus is more on differentiation… an open world setting more than city, driven by immersive daily living experiences… we’d rather take the time to get it right than ship something average in a rush” — is a reasonable answer to a crowded 2026 open-world calendar and should be read as a hypothesis, not a plan of record.

2.4 Corporate structure, and who owns it

NetEase, Inc. is a Cayman Islands holding company whose principal place of business moved from Hangzhou to Hong Kong during 2026. It is dual-listed — ADSs on Nasdaq (each representing five ordinary shares) and ordinary shares on the HKEX under 9999 — and on 2026-06-25 it completed conversion from a secondary to a dual-primary Hong Kong listing, a migration it was in fact required to make, because more than 55% of its worldwide trading by value had moved to Hong Kong and the exchange issued a Migration Exchange Notice under Listing Rule 19C.11.

Founder, CEO and director William Lei Ding beneficially owned approximately 45.5% of the outstanding shares as of 2026-02-28, held through Shining Globe International Limited → Shining Globe Holding Limited → the Shining Globe Trust, of which he is settlor with retained investment and dispositive power and of which he and his family are the beneficiaries. This is a controlled company in everything but name.

Because Chinese law forbids or restricts foreign investment in online-game operation, internet content provision and value-added telecoms, NetEase’s China-mainland operating businesses sit in variable-interest entities whose equity is held by Ding personally and by employees — Guangzhou NetEase Computer System and Hangzhou Yuedu are each 99.0% owned by Ding; Beijing NetEase Youdao Computer System is 71.1% Ding / 28.9% the Youdao CEO. Consolidation rests on contractual arrangements, not ownership. The VIEs generated 87.7%, 87.2% and 85.2% of total revenue in 2023, 2024 and 2025 while holding only 5.8% of consolidated assets. Investors in the ADSs are not purchasing equity in the entities that earn the money.

Verdict. This is a high-quality, cash-generative, asset-light business model — deferred-revenue funded, 1.8%-of-sales capital intensity, 70% of revenue self-authored — wrapped in a legal structure that gives shareholders a contractual rather than proprietary claim on 85% of it. Both halves of that sentence are load-bearing and this note returns to each.

3. Industry Dynamics

3.1 Size, growth, and the profit pool

China generated approximately US$44.8bn of games revenue in 2024, second globally only to the United States, with Niko Partners forecasting Chinese PC and mobile alone to exceed US$57bn by 2027 and the total market to pass US$100bn annually by 2030. Growth is mid-single-digit and driven by ARPU rather than users — the domestic player base is effectively saturated at over 600 million. That matters: a saturated market with a stable duopoly at the top is precisely where Greenwald would expect incumbent economics to be defensible, and where Marathon would expect returns to persist rather than mean-revert, provided new capital cannot easily enter.

3.2 The licence: the industry’s real barrier to entry

No game may be commercially operated in China without an ISBN approval from the National Press and Publication Administration. This is not a formality — it is the mechanism by which the state has twice frozen the industry’s supply of new product, in 2018 and again for nine months in 2021. It is also, in the ordinary course, the single hardest barrier to entry in the sector: it caps the number of titles the market can absorb and it advantages publishers with the relationships, compliance apparatus and portfolio breadth to keep a queue of approvals moving.

The regulatory environment today is the loosest it has been since 2017 and is still loosening. The NPPA approved 154 domestic and 4 imported titles on 2026-05-26, the fourth batch of the year, taking 2026 to date to 754 domestic and 25 imported approvals. Niko Partners reports approvals up 19% year to date with more than 2,300 titles expected to be approved in 2026. This is the most important industry fact in the memo, and it is the reason a sector that traded at trough multiples in 2022 can support a re-rating now.

What has not gone away: minors remain restricted to narrow play windows under the anti-addiction regime; the Minors Protection Law (last amended 2024-04-26) requires age classification, age-appropriate warnings and technical enforcement; and on 2023-12-22 the NPPA published draft rules capping in-game spending and prohibiting daily login rewards, a document that removed 16.06% of NetEase’s market value in one session on the heaviest volume of the five-year window before being withdrawn. The regulator has demonstrated both that it will intervene without warning and that it will retreat under economic pressure. Neither observation is comforting for a discount rate.

3.3 The distribution layer — where the profit pool is being redistributed

This is the section that explains the last two quarters, and it is the clearest live example of a Marathon capital cycle that this analyst has looked at recently.

Chinese Android app stores are not Google Play. They are operated by the handset manufacturers themselves — Huawei, OPPO, vivo, Xiaomi — which since forming the “Hardcore Alliance” in 2014 have jointly set terms at roughly 50:50 revenue sharing with game developers, versus the 30% that Apple and Google take in the rest of the world. Ding Lei attacked this “Android tax” publicly in 2021, arguing there was no reason Chinese channels should take 50% when foreign platforms take 30%. For a decade, the answer was that the channels controlled discovery and installation on the great majority of Chinese handsets, and the publishers had no alternative.

That is now breaking, from both ends simultaneously.

From the content side: Tencent pulled Dungeon & Fighter Mobile from Huawei, OPPO and vivo in June 2024 when revenue-share talks failed. NetEase has since late 2024 progressively withdrawn its major titles — including the Fantasy Westward Journey and Westward Journey franchises, Sword of Justice, Onmyoji, Minecraft and Sky: Children of the Light — from OPPO, vivo, Xiaomi and Huawei, with the OPPO shutdown reported on 2025-11-05 as a partnership expiry. Players are directed to NetEase’s own download channels instead.

From the regulatory side: China’s State Administration for Market Regulation opened an investigation into Apple’s App Store fees, and on 2026-03-15 Apple cut its China commission from 30% to 25% (and from 15% to 12% for Small Business Program participants), a change estimated to reduce Chinese developer costs by more than RMB6bn (US$870m) a year. Commentators have noted the obvious implication: the same regulatory logic points at the domestic Android channels next, whose take rate is twice Apple’s new one.

The economics of a successful boycott are enormous. Revenue that previously arrived net of a 50% channel fee now arrives net of nothing. But it is not free. The channels were also the marketing funnel, and self-distribution means buying the users the store used to supply — which is exactly what NetEase’s income statement shows, with selling and marketing expense rising from 9.4% to 11.2% of revenue in Q1-2026 (+27.7% in RMB). It is also a strategy available only to a publisher whose brand is strong enough that players will go and find the download themselves. There are precisely two such publishers in China.

3.4 Competitive intensity

Tencent is larger, better distributed (WeChat and QQ are the discovery layer NetEase does not have) and structurally advantaged in social/casual genres. NetEase’s answer has been depth rather than reach: the largest in-house R&D organisation in the country outside Tencent — 16,074 of 25,382 employees in games, 47.8% of total headcount classified as R&D — funding more than one hundred live titles and multiple simultaneous AAA cross-platform bets. Beneath the duopoly sit miHoYo (the Genshin Impact publisher, and the proof that a single studio can break in with one exceptional product), 37 Interactive, Lilith, ByteDance and 4399. Globally, NetEase now competes head-on with Western publishers on Steam and console, where it has no licence advantage and no distribution advantage — only product.

Verdict: structurally improving, from a genuinely bad base — but still a single-jurisdiction industry with a discretionary regulator. The licence regime is loosening after five years of tightening; the distribution rents that the handset makers extracted for a decade are being clawed back by the two players with the scale to fight and by the antitrust authority; and the demand curve is stable. Against that, this remains a hits business in which no structural feature protects a publisher from making a bad game, and a policy business in which no structural feature protects a shareholder from a Friday-afternoon NPPA circular. On Greenwald’s framework the industry has one true barrier to entry (the licence) and one durable scale economy (R&D and live-operations fixed costs); on Marathon’s, capital is currently leaving the distribution layer and returning to content, which is the direction an owner of content wants it to travel.

4. Competitive Position

4.1 Naming the moat

NetEase’s competitive advantage is, in Greenwald’s taxonomy, customer captivity reinforced by economies of scale, and it is real but narrower than the company’s marketing implies.

Customer captivity — the primary source. A player twenty-two years into Fantasy Westward Journey has accumulated character progression, guild relationships, in-game reputation and purchased items that do not transfer. This is habit plus sunk cost plus social lock-in, and the evidence that it binds is unusually strong because it is financial and multi-decade. The relevant test is Greenwald’s: if the captivity disappeared, which financial outcome would deteriorate? Two, immediately and observably — the twenty-year revenue annuity from the Westward Journey franchises, and the 69.7% games gross margin, which is only achievable because the marginal player is monetised without a marginal acquisition cost. Both are measurable. This clears the bar; it is a moat, not a story.

The corroborating operational evidence is specific and dated: 3.9 million peak concurrent users on FWJ PC in Q1-2026, a record for a title of that age; an 18-year-old MMO printing record quarterly revenue; a 2016 title at 200 million cumulative registrations celebrating its tenth anniversary with a TV-drama crossover; an 11-year-old title at its highest DAU since 2017. Games do not normally behave like this. Most of the industry’s catalogue value decays; NetEase’s oldest assets are re-accelerating.

Economies of scale — the secondary source. A 12,000-person games R&D organisation is a fixed cost that can only be carried by a company with a very large share of a very large market. It buys three things competitors cannot easily replicate: the ability to run 100+ titles as live services simultaneously (each requiring continuous content, events, balance patches and esports), the ability to fund several AAA cross-platform bets at once and survive the ones that fail, and the ability to keep an ISBN approval queue moving. Only Tencent matches it domestically.

What is NOT a moat here, despite being claimed. There is no network effect in the platform sense — NetEase does not own the social graph; Tencent does. There is no cost advantage in production; Chinese AAA development costs are converging on Western ones. There is no brand moat at the company level: players choose Marvel Rivals, not NetEase. And critically, the moat protects the installed base, not the pipeline. NetEase’s aggregate market share depends on hit rate, and hit rate is not defensible. This is why the note treats Ananta and Sea of Elements as options, not as forecasts.

4.2 The share-stability test

Greenwald’s diagnostic for a genuine barrier to entry is share stability over time. Applied at two levels, NetEase gives two different answers:

• At the franchise level: passes decisively. Titles operating for 10, 11, 18 and 22 years with flat-to-rising engagement is about as strong a share-stability result as exists in interactive entertainment.
• At the company level: fails. NetEase’s share of the Chinese market has moved with its release slate, and its 2023 revenue was flat while Tencent’s grew. A company whose share is stable only because its back catalogue is stable is a company whose future share depends on execution.

The honest characterisation is therefore: a wide moat around a large, ageing, extraordinarily durable installed base, and no moat at all around the next game.

4.3 The Blizzard round-trip — a bargaining-power experiment with a published result

NetEase’s relationship with Blizzard is the single cleanest natural experiment on its negotiating position, and it ran twice with opposite outcomes. The licences covering World of Warcraft, Hearthstone, Overwatch, Diablo and StarCraft in China expired in January 2023 and were not renewed — the games went dark, and licensed titles fell to 4.6% of NetEase’s revenue. Then, after Microsoft’s acquisition of Activision Blizzard, NetEase signed a new arrangement in 2024 covering those and other titles, and by FY2025 licensed games were 9.5% of revenue with Blizzard content achieving “record-high annual revenue” in China.

Read carefully, this cuts both ways. NetEase demonstrated it could walk away from a marquee foreign catalogue and survive — revenue grew in 2023 without it. But it also came back, on terms that the FY2025 20-F concedes are structurally inferior: royalty fees drove the 6.8% increase in games cost of revenues, and licensed arrangements “provide us with relatively less profit than games that we develop in-house.” The Minecraft licence from Mojang was renewed in 2023 and runs to 2028, at which point the same negotiation recurs. Licensed content is a top-line crutch with a margin cost and a counterparty-renewal risk, and the counterparty is Microsoft.

4.4 Direct comparison

Verdict: a durable advantage over a defined and large installed base, with no advantage whatsoever over the next release — and the market is pricing it as though only the second half were true. NetEase trades at the same multiple as Tencent, which has a genuine distribution network effect NetEase lacks, and at roughly half the multiple of Nintendo and Square Enix, whose franchises are not more durable than Fantasy Westward Journey and whose gross margins are 12–18 points lower. The gap is a jurisdiction discount, and it is defensible as a discount; the question the valuation section addresses is whether it is defensible at this size.

5. Growth History and Forward Opportunities

5.1 The record

(FY2024 and FY2025 operating profit are the sums of the four reported quarters; quarterly revenue sums foot exactly to the audited annual totals.)

The shape of this table is the whole growth story: revenue growth is unremarkable — 1.8% then 7.0% — while profit growth is not. Operating profit rose 21.1% in FY2025 on 7.0% more revenue, and 21.2% year-on-year in Q1-2026 on 6.1% more revenue. NetEase is not a growth company in the top-line sense. It is a company whose unit economics are improving faster than its volumes.

5.2 Where the growth came from

Self-developed titles. In-house revenue rose from RMB71.0bn (2023) to RMB72.6bn (2024) to RMB78.9bn (2025), +8.6% — driven, per the 20-F and the quarterly releases, by the Fantasy Westward Journey franchise, Where Winds Meet and Marvel Rivals. This is the highest-quality growth in the business: organic, self-authored, margin-accretive, and carrying no royalty.

Licensed titles. From 4.6% of revenue (2023) to 7.4% (2024) to 9.5% (2025) as the Blizzard catalogue returned. This is genuine revenue and it delivered “record-high annual revenue” for Blizzard titles in China in 2025, but it is lower-margin, renewal-dependent, and now largely lapped.

Global. This is the newest and most important vector. Marvel Rivals reached 40 million registered players in two months; Where Winds Meet went from 15 million to 80 million+ cumulative players in a year, ranked #2 on Steam globally on a content update, and holds a 78% positive Steam rating — in markets management itself calls “traditionally difficult for Chinese [titles] to penetrate.” Sword of Justice launched globally in Q4-2025 to what management described as a strong reception. NetEase does not disclose overseas revenue as a percentage of total, which is a material disclosure gap given that overseas expansion is the primary equity story — it is recorded in the Open Questions section.

What shrank. Cloud Music revenue fell 2.4% in FY2025 as social-entertainment revenue declined RMB830.6m, partially offset by RMB639.9m of growth in online music subscriptions — a mix shift toward the better business, which is why its gross margin rose from 33.7% to 35.7%. Innovative businesses fell 15.9%. Youdao grew 5.0% but its gross margin fell from 48.9% to 44.3% as low-margin online marketing displaced learning services.

5.3 Forward opportunities, ranked by evidence

1. Margin, not volume. The Android disintermediation is worth more to earnings over the next two years than any plausible revenue acceleration. A 6-point games gross-margin gain on a RMB92bn segment is RMB5.5bn of gross profit — roughly 15% of FY2025 operating profit — before the Apple cut, which is present in only two weeks of Q1-2026 and is fully in from Q2-2026.
2. Sustained global live-service revenue. Where Winds Meet and Marvel Rivals are not launches any more; they are annuities in their second year, with content roadmaps (a Marvel “year-long road map… leading into the release of Avengers: Doomsday in December”) that tie monetisation to external cultural events. This is the first time NetEase has had Western live-service assets rather than Western launches.
3. The pipeline. Sea of Elements targets a Q3-2026 window; Ananta is undated. Both are genuine options with genuine option value and zero contractual certainty. They should be underwritten at close to nothing.
4. The regulatory tailwind. Approvals up 19% year to date with 2,300+ expected in 2026 lowers the friction on every future release.
5. Stock Connect. The dual-primary conversion completed 2026-06-25 makes NetEase eligible for southbound Stock Connect inclusion — a flow event, not a fundamental one, but a real one for a stock whose factor profile shows it is currently owned as a China ADR.

Verdict: high-quality growth of an unusual kind. It is not volume growth — revenue compounds at mid-single digits and will likely continue to. It is margin and mix growth, organic, self-funded, and driven by the company reclaiming economics it was previously paying away. That is a better kind of growth than the reverse, because it does not require capital and does not require the market to expand. The risk is precisely symmetric: margin growth of this kind is finite. You cannot take the channel fee to zero twice.

6. Financial Quality

6.1 The margin bridge — the most important table in this note

The single most striking line in NetEase’s financial history is Q1-2026’s cost of revenues: RMB9,374.2m against RMB10,349.1m a year earlier — down 9.4% in absolute terms while revenue rose 6.1%. In a business whose largest cost of revenue is a variable revenue-share on gross billings, absolute cost cannot fall on rising revenue unless the rate has changed. It has. Management gave the segment detail on the Q1-2026 call: games-segment gross margin 74.8%, versus 68.8% a year earlier, “primarily driven by lower revenue sharing costs related to the platform.”

Three corroborations make this a supported finding rather than a management claim:

1. The same driver appears one quarter earlier. The Q4-2025 release attributes the year-on-year cost decline to “lower royalties for licensed games and lower revenue-sharing costs.” The trend started in Q4-2025 and accelerated.
2. The independent trade record. NetEase’s withdrawal of major franchises from the OPPO, vivo, Xiaomi and Huawei app stores has been reported continuously since late 2024, most recently the OPPO shutdown on 2025-11-05.
3. The offsetting cost shows up where theory says it should. Selling and marketing expense rose 27.7% in RMB, from 9.4% to 11.2% of revenue — precisely the user-acquisition spend that a publisher must fund when it stops paying an app store to supply installs.

The bridge, in points of revenue, Q1-2026 versus Q1-2025: gross margin +5.3, S&M −1.8, G&A +1.2, R&D +0.5, net operating margin +5.2. Roughly one third of the gross-margin windfall was handed straight back in marketing; the remainder was kept, augmented by operating leverage on G&A and R&D.

The honest caveat. The G&A line fell 33.4% in RMB year-on-year (RMB956.3m → RMB636.6m), which is a very large move for a corporate overhead line and was given no explanation by the company. If some of it is a one-time credit or reversal, roughly 1.2 points of the 5.2-point operating-margin gain is not repeatable. Recorded as an open question, not netted out.

6.2 What the margin story is worth, and what happens if it reverses

• If Q1-2026 is the new run rate: annualising gives operating profit of RMB50.6bn (US$7.34bn) against a TTM RMB38.1bn — a 33% step-up.
• If the games gross margin fully reverts to the Q4-2025 level of 64.2% total: Q1-2026 gross profit would have been RMB19,640m rather than RMB21,217m, cutting operating profit by RMB1,577m and TTM operating profit to RMB36,474m (US$5,288m). Even a complete reversal leaves the business at 11.5x EV/EBIT.
• The Apple cut is additive and not yet in the numbers. The 30%→25% reduction took effect 2026-03-15, covering two weeks of Q1-2026 and all of Q2-2026 onward. NetEase does not disclose its iOS mix, so the amount is not calculable from public sources — but the direction and the timing are certain, and the sign is positive.

6.3 Margin structure over five years

This is a business whose gross margin has risen 11.4 points in five years and operating margin 12.1 points — before Q1-2026. The FY2020–22 period was depressed by the low-margin e-commerce and live-streaming businesses that are now being deliberately shrunk, and by higher channel costs. The FY2023 step-up coincides with the mix shift into higher-margin self-developed PC titles. The economics do improve with scale, and they have been improving for five consecutive years. That answers the financial-quality question directly and affirmatively.

The effective tax rate deserves a note: it fell from ~20% to ~14–15% in FY2023–25 because several NetEase entities qualify as High and New Technology Enterprises, which carry a reduced 15% PRC rate. Q1-2026’s 18.9% is a step up and management describes the rate as varying with the entity mix each quarter. The tax rate is a real and unhedged variable: a return to 20% would cost roughly 6% of net income.

6.4 Cash generation and its quality

FY2025 operating cash flow was RMB50.7bn (US$7,256m) against net income of RMB34.8bn — a 1.50x conversion ratio. The 20-F’s own bridge attributes the gap to non-cash add-backs (share-based compensation RMB3.6bn, investment impairments RMB2.7bn, D&A RMB2.2bn, equity-method results RMB1.5bn, unrealised FX RMB0.7bn), offset by deferred taxes and fair-value gains, plus a working-capital contribution of US$1,247m.

Applying the standard quality-of-earnings test — when reported operating cash flow exceeds net income by 50%, find the working capital — Here it is not hidden and it is not abusive — it is the contract-liability build, i.e. players prepaying. Short-term deferred revenue rose from US$2,691m (Dec-24) to US$3,787m (Dec-25), +41%, and contract liabilities rose a further RMB1,384.4m in Q1-2026. But it must be stripped before a run-rate FCF number is used:

Reported FCF is an 8.2% yield on market capitalisation and 11.5% on enterprise value; working-capital-neutral FCF is 6.7% and 9.4%. Both are high. Neither is manufactured — there is no customer prepayment with an offsetting capital commitment buried in “other investing,” and capital expenditure is not understating a replacement obligation: capex of US$294m sits against D&A of US$321m, a 0.91x ratio on gross PP&E of US$2,354m. This business has no undisclosed future call on cash.

6.5 Where GAAP earnings mislead — in both directions

NetEase’s reported net income is materially noisier than its operating profit, because of a US$3.07bn long-term investment book and a Cayman-parent/RMB-operations currency structure.

• Q4-2025: operating profit RMB8,319.2m, but net income attributable only RMB6,242.0m, after an investment loss of RMB1,669.1m and an FX loss of RMB518.0m.
• Q1-2026: investment income collapsed from RMB692.8m to RMB5.5m and FX losses were RMB622.1m — yet net income still rose 3.6% year-on-year, because operations more than covered it.
• FY2025 impairment losses on investments were RMB2.7bn (US$389.6m), against RMB1.3bn in FY2024 and RMB0.5bn in FY2023 — an escalating three-year series.

Interpretation: read operating profit, not net income, for this company. The Q4-2025 GAAP print is a large part of why the stock de-rated 30% between September 2025 and March 2026, and the operating business did not deteriorate at all in that window. Equally, in a quarter where marks go the other way, GAAP net income will overstate. The non-GAAP measure NetEase publishes adds back only share-based compensation and does not strip the investment marks, so it does not solve the problem either.

6.6 Balance sheet and invested capital

At 2026-03-31 (RMB m): cash and equivalents 50,815.2; current time deposits 80,109.9; restricted cash 4,503.0; short-term investments 39,978.7; non-current time deposits 3,045.0 = 178,451.8 gross liquid assets; less short-term loans 10,955.5 = net cash RMB167,496.3m (US$24,282m), against RMB163,540.7m at December 2025. Shareholders’ equity RMB164,722.2m; total assets RMB231,397.0m; total liabilities RMB62,047.6m of which contract liabilities are RMB21,811.5m.

Two structural observations follow.

First, the company has no leverage in any economic sense, but it does carry RMB11.0bn of short-term borrowings — up from RMB6.4bn a quarter earlier — against RMB178bn of liquid assets. Onshore RMB borrowing against offshore USD deposits is a standard cash-management and currency-hedging construct for a China-based issuer with a Cayman parent. It is also the reason vendor “net debt” fields that net only cash against debt are meaningless for this ticker.

Second, operating invested capital is negative. Total assets of RMB231.4bn less liquid assets (RMB178.5bn) less long-term investments (RMB~22.5bn in other long-term assets) leaves an operating asset base of roughly RMB30bn, against non-debt operating liabilities of RMB46.5bn (payables, salary, taxes, contract liabilities, accruals). The games business is funded by its customers and its suppliers and consumes no shareholder capital at all. The reported ROIC of 18.7% is therefore a cash-diluted number and dramatically understates the operating business; the correct statement is that return on operating capital is not computable because the denominator is below zero.

Verdict: the economics unambiguously improve with scale, and the improvement has been continuous for five years. Gross margin +11.4 points and operating margin +12.1 points over 2020–2025, before a further 5.2-point step in Q1-2026; 1.8%-of-sales capital intensity; customer-funded working capital; 0.91x capex-to-D&A with no hidden replacement obligation; and a genuinely fortress balance sheet. The two blemishes are real but second-order: a venture portfolio impairing at an accelerating rate, and a GAAP net-income line that is a poor guide to the business in either direction.

7. Capital Allocation

This is where a strong report becomes a qualified one.

7.1 The record

7.2 What is good

M&A discipline is exemplary, and this deserves to be said plainly. In each of the last five years NetEase spent less on acquisitions than it received from divestitures — it has been a net seller of businesses every year since 2021. Goodwill is US$265m, 0.8% of total assets. There is no Zynga here, no US$5.9bn impairment, no serial-acquirer dilution. Set against Take-Two’s record (a US$9.7bn acquisition followed by roughly US$5.9bn of goodwill impairment and heavy dilution), NetEase’s approach is the correct one for a company whose competitive advantage is internal development capability. It builds rather than buys, and the build has produced Marvel Rivals and Where Winds Meet.

Compensation is not a leak. Aggregate 2025 cash compensation for all executive officers and directors in office was RMB44.5m (US$6.4m), and RSU grants to executive officers would, if fully vested, represent less than 1% of outstanding ordinary shares. For a US$85bn company that is close to nominal. Share-based compensation of US$521.6m is 3.2% of revenue — against Roblox at about 23% of revenue, this is a company that pays its people in cash and keeps the equity.

Reinvestment in the business is real and rising. R&D expense ran at RMB4,482m in Q1-2026, 14.7% of revenue, funding 12,000-plus R&D staff. That is the correct place for the money in a business whose moat is development capability, and it is expensed, not capitalised.

7.3 What is not good

The buyback has been abandoned at precisely the wrong moment. Repurchases fell from US$1,210m in FY2024 to US$91m in FY2025 — a 92.4% cut — in a year when free cash flow rose 36% and the stock spent the first half of the year below $105. Cumulative usage of the current US$5.0bn authorisation is US$2.1bn (23.2m ADSs) over 3.2 years, or 42%. On 2025-11-20 the board extended the authorisation by a further 36 months to January 2029, which is the corporate-governance equivalent of renewing a gym membership you do not use.

The share count is rising. Ordinary shares outstanding went from 3,167,959,016 (Dec-2024) to 3,192,111,251 (Dec-2025), +0.76%, and the Q1-2026 weighted diluted count is higher again at 3,227,325k. With US$521.6m of annual share-based compensation and US$91m of repurchase, the buyback offset roughly 17% of the equity issued to employees. Shareholders are being modestly diluted by a company sitting on US$24.3bn of net cash.

The cash pile is now large enough to be a capital-allocation decision in itself. Net cash is 28.5% of market capitalisation and grew RMB4.0bn in Q1-2026 alone after paying a US$1.16-per-ADS dividend. The stated return on it is falling — interest income declined from RMB1,060.9m (Q1-2025) to RMB890.3m (Q1-2026), −16%, roughly 2.2% annualised, as Chinese rates decline. Against a business earning an unbounded return on operating capital and a stock at 12.4x cash-adjusted earnings, every incremental dollar held in time deposits is a dollar earning 2.2% instead of retiring an 8% earnings yield. There is no disclosed acquisition programme, no disclosed capacity project, and no disclosed reason for the hoard.

The venture book is a small, persistent leak. A US$3.07bn long-term investment portfolio has generated impairments of RMB0.5bn / 1.3bn / 2.7bn in 2023 / 2024 / 2025 — RMB4.5bn (US$0.64bn) cumulative and accelerating. That is a poor return on 5% of the balance sheet. It does not change the thesis; it does mean the “excess” cash has not, historically, been well deployed when it has been deployed.

7.4 The governance frame

Ding owns 45.5%. That aligns him with minority shareholders on price, and it explains both the low compensation and the absence of empire-building M&A — this is a founder spending his own money. It also means there is no mechanism by which shareholders can force a change in the distribution policy. As an FPI, NetEase does not publish a US-style compensation discussion and analysis with a performance-metric table, so the SHAK/HBM-style audit of whether incentives contain a per-share or return-on-capital metric cannot be run on public disclosure. That is recorded as a disclosure limitation, not as a finding.

The distribution policy is a genuine disagreement, not a scandal: the dividend has more than tripled in four years (US$552m → US$1,977m) and the trailing yield is 2.27%, which is not nothing. But a company returning 30% of free cash flow while trading at 12.4x cash-adjusted earnings is making an implicit statement that it has better uses for the other 70%, and it has not said what they are.

Verdict: management has allocated capital intelligently but not fully. On the two ways capital allocation usually destroys value — bad acquisitions and excessive equity issuance — NetEase scores near the top of any comparable set. On the third — failing to return capital that has no productive use — it scores badly and is getting worse. The distinction matters for the thesis: this is a company where the risk is opportunity cost, not impairment. That is a materially better problem to have than the alternative, and it is also the most easily fixed. It is the reason Claude’s Take names the buyback as the single upgrade trigger.

8. Changes and Headwinds — Last Two Years

Blizzard returns (2024). After the January 2023 licence lapse took NetEase’s licensed-game revenue to 4.6% of total, a new arrangement with Blizzard under Microsoft ownership restored World of Warcraft, Hearthstone, Overwatch, Diablo and StarCraft to China. Licensed games reached 9.5% of FY2025 revenue and Blizzard titles achieved “record-high annual revenue” in China in 2025, with the Midnight expansion launching in China simultaneously with the global release on 2026-03-03 and Diablo IV’s expansion launching simultaneously in April 2026. Thesis effect: strengthens the top line, dilutes the margin, adds a renewal dependency on Microsoft.

Two genuine global hits (Dec 2024 – 2025). Marvel Rivals (40m registered players in two months) and Where Winds Meet (15m → 80m+ players in a year, #2 on Steam globally, 78% positive rating) are the first NetEase titles to succeed in Western markets at scale. Sword of Justice launched globally in Q4-2025. Thesis effect: materially strengthens — it converts “China games company” into “global games company with a China base,” which is the difference between the current multiple and a Western one.

The Android channel revolt (late 2024 → present). NetEase progressively withdrew its largest franchises from the OPPO, vivo, Xiaomi and Huawei stores. Games gross margin: 68.7% (FY24) → 69.7% (FY25) → 74.8% (Q1-26). Thesis effect: this is the single largest positive change in the file, and the most uncertain in duration.

Apple cuts China commission 30% → 25%, effective 2026-03-15, under pressure from the State Administration for Market Regulation; estimated to save Chinese developers more than RMB6bn a year in aggregate. Thesis effect: additive, quantifiable in direction only, and almost entirely still ahead of the reported numbers.

Section 16 arrives (2026-03-18). The Holding Foreign Insiders Accountable Act, enacted 2025-12-18 in the FY2026 NDAA, extended Section 16(a) to directors and executive officers of foreign private issuers. On 2026-03-17, eight NetEase insiders filed initial Form 3s; the first-ever Form 4 followed on 2026-06-30. Thesis effect: neutral to mildly positive — for the first time, US investors get real-time insider-transaction visibility on this issuer.

Dual-primary Hong Kong listing (2026-06-25). Following a Migration Exchange Notice under HK Listing Rule 19C.11 — issued because more than 55% of worldwide trading by value had migrated to Hong Kong — NetEase converted from a secondary to a dual-primary HKEX listing, announced on 2026-03-02 and completed 2026-06-25, and moved its principal place of business to Hong Kong in April 2026. A Second Amended and Restated 2019 Share Incentive Plan was filed on 2026-07-15 in connection with the conversion. Thesis effect: strengthens — it makes NetEase eligible for southbound Stock Connect, largely defuses the ADR-delisting tail, and the market agreed immediately (+7.7% and +4.4% on the two sessions following). The cost is higher compliance burden: NetEase loses the secondary-listing waivers and becomes subject to the full Hong Kong Listing Rules and the Takeovers Code.

SEC comment-letter cycle (Jul–Aug 2025). The Office of Technology required NetEase to (i) stop excluding Hong Kong and Macau from its definition of “China” for risk-disclosure purposes and (ii) restore risk language it had narrowed, specifically the sentence that “The PRC government may also intervene with or influence our operations as it deems appropriate to further regulatory, political and societal goals at any time.” NetEase agreed and the sentence appears in the FY2025 20-F. Thesis effect: neutral in substance, mildly negative on disclosure posture — management’s instinct was to soften.

Non-games businesses continue to shrink or stall. Cloud Music revenue −2.4% in FY2025; innovative businesses −15.9%; Youdao +5.0% but with gross margin down 4.6 points as low-margin advertising displaced learning services. Thesis effect: mildly positive on margin mix, neutral on value — these are increasingly rounding errors.

The GAAP-quality air pocket (Q3-2025 – Q4-2025). Operating margin fell back to 28.3% and 30.2%, and Q4-2025 net income dropped to RMB6.2bn on a RMB1.67bn investment loss and a RMB0.52bn FX loss. The stock de-rated 30% from the September 2025 high into late March 2026. Thesis effect: no operating deterioration occurred; the episode is best read as a demonstration of how much of NetEase’s reported earnings volatility is non-operating.

Verdict: the last two years have strengthened the thesis, and the strengthening is concentrated in the last three quarters. Two global hits, a structural margin reset, a regulatory thaw, and a listing change that removes the tail risk that defined the 2022 trough. The offsetting developments — a Microsoft dependency, a rising compliance burden, and the SEC episode — are real but small. The single genuine unknown is the durability of the channel-fee gain.

9. Risk Analysis

The two risks that actually matter are #1 and #2, and they are not independent. NetEase’s earnings power is currently benefiting from a redistribution of profit away from the handset makers, which the antitrust regulator has tacitly endorsed by squeezing Apple. A Chinese regulator willing to redistribute rents toward content owners is also a Chinese regulator willing to redistribute them away. The same discretionary authority is on both sides of the ledger. That is the deepest reason this business does not deserve a Western multiple — and, given the size of the discount currently applied, it is also not a reason to avoid it.

10. Valuation Discussion

10.1 A necessary preliminary: every vendor’s headline multiple for this ticker is wrong

Before any valuation work, the data must be rebuilt. ROIC.ai reports FY2025 enterprise value of US$417.7bn and price-to-book of 19.16x against an actual market capitalisation of ~US$85bn and price-to-book of ~3.6x. The diagnosis is exact: ROIC multiplies the ADS price by the ordinary-share count — a clean 5x error, because one ADS represents five ordinary shares. Its P/E (18.17x) and P/S (5.45x) are correct because both numerator and denominator are per-ADS; every EV-derived and book-derived multiple is inflated exactly fivefold. yfinance is wrong in the opposite direction, reporting price-to-sales of 0.746 and enterprise value of US$270.9bn by comparing an RMB revenue figure to a USD market capitalisation. Neither vendor’s EV, P/B, EV/EBITDA or EV/EBIT may be used for NTES. Everything below is hand-built from the filed statements.

10.2 The capital structure, rebuilt

The net-cash figure is the company’s own published definition — total cash and equivalents, current and non-current time deposits, restricted cash and short-term investments, less short- and long-term loans — and it reproduces exactly from the Q1-2026 balance sheet.

10.3 Multiples

The cash adjustment is done properly rather than crudely: TTM interest income of roughly RMB3,719m (US$539m) is stripped from earnings at the same time as the cash is stripped from the price, at the 17% blended effective tax rate — US$0.69 per ADS. Failing to do that would double-count the cash.

10.4 Own-history percentiles — and why they disagree

AZI’s own-history valuation index (dated 2026-07-24, at $119.55) reports: composite 37.7th percentile; P/E 20.6th; P/B 25.2nd; P/S 67.4th, with all three components valid (n_components = 3, no negative book equity, no GAAP-distorting valuation-allowance release, no REIT-style earnings artifact).

The split is the analysis. NetEase sits near the cheapest quintile of its own decade on earnings and book, and in the expensive third on sales — because the net margin has gone from 16.4% (FY2020) to 30.0% (FY2025) to 34.9% (Q1-2026). Price-to-sales has risen because each unit of revenue now carries more than twice the profit it did five years ago.

Which percentile is the true reading depends entirely on whether the margin is durable. This is the mirror image of the classic low-margin restaurant case, where a 2nd-percentile price-to-sales concealed a 54x earnings multiple on a thin margin. Here a 67th-percentile price-to-sales conceals a 21st-percentile earnings multiple on a margin that five consecutive years of evidence say is structural, not cyclical. The note’s judgement — labelled as judgement — is that the P/E percentile is the better guide, because the margin expansion has a documented mechanism (channel disintermediation plus mix shift out of low-margin e-commerce and live streaming) rather than a cyclical one.

10.5 Peer comparison

NetEase trades in line with Tencent — a company with a genuine distribution network effect it lacks — and at roughly half the multiple of Nintendo and Square Enix, whose franchises are no more durable than Fantasy Westward Journey and whose gross margins are 12–18 points lower. It trades far below Take-Two despite being vastly more diversified. The entire gap is a China discount, and it is applied to earnings that are increasingly generated outside China.

10.6 Embedded expectations

At an enterprise value of US$60.8bn, TTM operating profit of US$5,517m taxed at 17% gives NOPAT of US$4,579m — a 7.5% after-tax operating earnings yield. Capitalising that at a 9% cost of equity (a defensible rate for a China-domiciled, VIE-structured asset — 5–6% would be the Western games comparable) gives a no-growth perpetuity value of US$50.9bn, which is 84% of today’s enterprise value.

The market is therefore paying for roughly 1.5% of perpetual real growth in NetEase’s operating profit. For context, operating profit grew 21.1% in FY2025 and 21.2% year-on-year in Q1-2026; revenue has grown 7.0%; and the industry is expected to compound mid-single-digit with a licensing regime that is loosening. The embedded expectation is not merely conservative — it prices a business that grows profit at roughly the rate of inflation, forever, starting from a quarter in which it grew profit twenty-one percent.

That is the bull case stated as arithmetic. The counter-argument is equally clean: a 9% cost of equity may be too low for an asset where 85% of revenue is earned by entities the shareholder does not own and where the regulator has twice erased more than a tenth of the market value in a session. At an 11% discount rate, the no-growth perpetuity is US$41.6bn and the market is underwriting ~3.5% perpetual growth — still undemanding, but no longer a giveaway. The valuation debate here is entirely a discount-rate debate, not a forecast debate, and readers should form their own view on the appropriate China risk premium.

10.7 Scenarios

ASSUMPTIONS, stated explicitly: all three scenarios assume no change in VIE enforceability, no ADR delisting or US investment restriction, no revival of in-game spending caps, and no reversal of the current channel-fee position. Net cash is grown at the scenario’s own free cash flow less the current ~30% payout, except in the bull case where a real buyback is assumed. Exit multiples are judgement, not derived.

The point of the table is the asymmetry, not the base case. The bear scenario — which assumes the entire Q1-2026 margin achievement reverses and growth halves — produces an outcome roughly 9% below the current price over three years, because 28.5% of the price is cash that the bear case does not destroy. The base case is +39% and the bull +78%. A distribution in which the downside case is a mid-single-digit annualised loss and the central case is a low-teens annualised gain is an unusual one to find at 17x trailing earnings.

Verdict: the market is correctly pricing NetEase’s jurisdiction and incorrectly pricing its business. What it has right: the VIE structure is a genuine impairment of the shareholder claim, the regulator is genuinely discretionary, and management genuinely will not return capital. What it has wrong: it is applying a 1.5%-perpetual-growth valuation to a franchise portfolio with twenty-year revenue annuities, a documented and still-incomplete structural margin reset, two new global hits in their second year, and negative operating invested capital — and it is doing so while classifying the security, per the factor model, alongside nine other Chinese ADRs and not one games company.

11. Variant Perception

11.1 What consensus believes

The consensus view of NetEase is that it is a cheap, well-run, structurally-capped Chinese internet asset: a good operator whose earnings are real but whose multiple can never expand because the jurisdiction caps it, whose growth is mid-single-digit, and whose cash pile is a permanently trapped asset. On this view 17x trailing is roughly fair — you own it for the earnings, not for a re-rating, and you accept that a Beijing headline can take 15% out of it at any time.

The factor evidence says this view is not merely held but institutionalised. NetEase’s largest loading in the full model is Industry: Social Media at +1.307, followed by Country: China at +0.708; market beta is only 0.39–0.48. Its ten nearest factor-neighbours are TME, TAL, YMM, BIDU, BILI, WB, VIPS, ATHM, QFIN and PDD — every one a Chinese ADR, not one a games company. The security is owned as China exposure. Its business results are increasingly not China exposure.

11.2 The strongest bull case

1. You are paying 12.4x cash-adjusted non-GAAP earnings for a 33%-operating-margin business with negative invested capital, a 65.7% gross margin, and a 22-year revenue annuity at its core.
2. The margin reset has a mechanism and is not finished. The channel disintermediation drove games gross margin from 68.8% to 74.8% in four quarters; Apple’s 30%→25% China cut landed on 2026-03-15 and is in only two weeks of the last reported quarter.
3. The global franchise is real for the first time. Marvel Rivals at 40m players in two months and Where Winds Meet at 80m+ with a 78% Steam rating and a #2 global top-seller ranking are not exports — they are Western live-service assets. Each incremental dollar earned outside China weakens the case for the China discount.
4. The regulatory backdrop is at its most permissive since 2017 — 754 domestic approvals year-to-date, up 19%, with 2,300+ expected in 2026.
5. The delisting tail has been substantially defused by the 2026-06-25 dual-primary Hong Kong listing, which also opens southbound Stock Connect eligibility.
6. The founder owns 45.5% and has never sold a disclosed share. Total disclosed insider selling across five years is US$9.5m — 0.011% of market capitalisation.
7. The embedded expectation is ~1.5% perpetual growth against 21% actual operating-profit growth.

11.3 The strongest bear case

1. You do not own the business. 85.2% of revenue is generated by variable-interest entities in which NetEase holds no equity, whose registered equity is held personally by the CEO, and whose consolidation rests on contracts of untested enforceability under PRC law.
2. The margin gain is a negotiating position, not a contract. No agreement underpins the 74.8% games gross margin. If the handset makers settle at 30%, NetEase will likely return to their stores — recovering reach, giving back margin, and taking the P/S percentile’s side of the argument.
3. Management will not return the capital. Buybacks fell 92% to US$91m; the share count rose; a US$5bn authorisation is 42% used after 3.2 years; interest income on the hoard is falling 16% a year. A discount with no forcing mechanism can persist indefinitely — this is precisely the pattern seen at other cash-backed China-listed assets, which have stayed cheap for years.
4. The regulator that giveth can taketh. The same discretionary authority that is squeezing Apple’s take rate published draft in-game spending caps on 2023-12-22 and removed 16.06% of NetEase’s value in a session.
5. This is a hits business at the margin. The back catalogue is defensible; the next game is not. Ananta is undated into a crowded 2026 open-world calendar that management itself acknowledges.
6. The five-year record is genuinely poor: 7.3% a year, 44% volatility, −51.4% drawdown, 0.12 Sharpe. Investors who bought the cheapness in 2021 waited four years to get back to flat.

11.4 The 3–5 assumptions that actually matter

11.5 Where consensus is most likely offsides

The factor evidence and the fundamental evidence point at the same place. Consensus prices NetEase as China beta; the model confirms it by grouping it with nine Chinese ADRs and zero games companies; and the negative Momentum loading (−0.142) into a regime where Value is +16.1% over twelve months (z +1.88) and the social-media basket NetEase loads on hardest is −13.1% over the same window confirms that this is an out-of-favour, not a crowded, position.

The variant perception is therefore precise: NetEase’s margin structure is being re-rated by a supply-side change in its own industry’s distribution layer, and the market is still valuing it on the jurisdiction. The specific, checkable claim is that the games gross margin holds above 72% through FY2026 while the market continues to apply a China multiple to it. If that happens, the earnings come through and the multiple does not need to expand for the outcome to be good; if the multiple expands too, the base case is conservative.

12. Fact vs. Interpretation

13. Open Questions

1. How much of the Q1-2026 games gross margin is channel mix versus a one-off? The company gave one sentence. The specific unknown is what share of China mobile revenue now flows through NetEase’s own distribution versus the handset stores. Resolvable only by disclosure or by watching two more quarters.
2. What is the overseas revenue percentage? NetEase does not disclose it. For a company whose equity story is global expansion and whose valuation discount is jurisdictional, this is the single most conspicuous disclosure gap in the file, and it should be the first question on any management call.
3. Why did G&A fall 33.4% year-on-year in Q1-2026? RMB956.3m → RMB636.6m is worth 1.2 points of operating margin and was left unexplained.
4. What are the Blizzard licence terms and renewal date? The FY2025 20-F discloses the Minecraft licence runs to 2028 but gives no term for the 2024 Blizzard arrangement, which now underpins 9.5% of revenue.
5. What is the intended use of US$24.3bn of net cash? No acquisition programme, no capacity project, no stated target capital structure, and a buyback authorisation that is being extended rather than used.
6. What does the Second Amended and Restated 2019 Share Incentive Plan (filed 2026-07-15) change? Pool size and vesting terms not analysed; relevant given the share count is already rising.
7. What is in the 13F portfolio? NetEase files Form 13F-HR quarterly. The holdings were not itemised for this note.
8. Is there a succession plan? Ding is founder, CEO, 45.5% owner, sole director of the entity of record for his stake, and the personal registered equity holder of the VIEs. Nothing is disclosed.
9. What compensation metrics does the board actually use? As an FPI, NetEase publishes no US-style CD&A metric table, so the standard proxy audit cannot be run.
10. What would a paid industry primer add? No subscription sell-side sector research was consulted for this note; the industry framing rests entirely on public sources, company filings and trade press, and would benefit from a cross-check against a specialist China-games data provider.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

The cleanest single test in this note is bull #1 / bear #1: NetEase discloses its games-segment gross margin on every quarterly call, and the next print is due 2026-08-13. One number, published quarterly, resolves the central disagreement.

15. Source Appendix

See NTES_source_appendix.md (Appendix B of the combined report) for the full source list with URLs and access dates.

APPENDIX A — Standard Diligence Questionnaire

NetEase, Inc. (NASDAQ: NTES / HKEX: 9999) — as of 2026-08-01

A supplemental due-diligence questionnaire accompanying the note above. Labels: FACT / INTERPRETATION / ASSUMPTION where the distinction matters.

General

What thoughtful questions have other investors asked about this company?

Four recur, and the note answers each.

1. “Is the cash real, and can it ever be got at?” — FACT: RMB167.5bn (US$24.3bn) at 2026-03-31, reproducible line-by-line from the filed balance sheet, and the company pays a growing US-dollar dividend out of it (US$1,977m in FY2025, up from US$552m in FY2021), so cash does cross the border. INTERPRETATION: the constraint is willingness, not capability.
2. “Doesn’t the VIE structure make the equity worthless in a bad state of the world?” — FACT: 85.2% of FY2025 revenue is earned by VIEs whose registered equity is held personally by the CEO. INTERPRETATION: in a genuine adverse-enforcement scenario, yes, the claim is impaired; this is the single largest justification for the discount, and it is why the note argues about the size of the discount rather than its existence.
3. “Is NetEase just a Tencent tracker?” — FACT: FactorsToday classifies it with nine Chinese ADRs and zero games companies; its biggest loading is a China-internet/social-media basket (+1.31). INTERPRETATION: it is currently owned as one, which is precisely the mispricing the note identifies.
4. “What happens when the game pipeline is empty?” — FACT: the catalogue includes titles at 10, 11, 18 and 22 years of operation with record engagement metrics in Q1-2026. INTERPRETATION: the back catalogue is far more durable than the industry norm, but new-title share is not defensible; the honest answer is that the pipeline should be underwritten at close to zero, which the note’s base case does.

A fifth question is one investors should be asking and largely are not: what share of revenue now comes from outside China? NetEase does not disclose it. Given that the global titles are the reason the business is changing and the China domicile is the reason the multiple is low, this is the most valuable undisclosed number in the file.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low?

INTERPRETATION: at a structural high that is not obviously cyclical. Q1-2026 operating margin of 41.4% is the highest in the company’s history and is 5.2 points above the prior-year quarter and 11.2 points above the Q1-2024 level. But the driver is not a demand cycle — revenue grew only 6.1%. It is a change in the split of revenue between NetEase and its distribution channels. That is a level shift, not a cycle, unless the channels win the negotiation back.

The honest framing: gross margin is at a structural high; volumes are at a normal level; the venture-portfolio marks embedded in GAAP net income are at a cyclical low (FY2025 impairments of RMB2.7bn were triple FY2023’s).

Driven by the external environment or internal actions?

Internal, unusually so. The margin gain came from a deliberate decision to withdraw titles from the Android app stores and absorb the user-acquisition cost — visible in S&M rising from 9.4% to 11.2% of revenue. The externally-driven component (Apple’s 30%→25% cut, effective 2026-03-15) is additive and mostly still ahead.

How stable are revenues?

Very, by industry standards. Quarterly revenue over the last nine quarters has ranged RMB25.5bn–30.6bn — a peak-to-trough spread of 20%, most of which is seasonal (Q3 benefits from “a high level of diversified events for the summer period,” per the Q4-2025 release). No quarter in the window declined more than 4.6% sequentially. The reason is the deferred-revenue model plus a hundred-title portfolio: no single game is large enough to swing the consolidated result.

Outlook for products/services?

Established franchises: stable to modestly growing, with genuine re-acceleration evidence (FWJ PC at a record 3.9m peak concurrent users; an 18-year MMO at record quarterly revenue). Global titles: Where Winds Meet and Marvel Rivals are in their second year as live services rather than launches. Pipeline: Sea of Elements targeted for Q3-2026, Ananta undated. Non-games: Cloud Music −2.4%, innovative businesses −15.9%, Youdao +5.0% with falling margin — collectively a drag being managed down.

How big will this market be — growing, shrinking, domestic or international?

FACT: China generated ~US$44.8bn of games revenue in 2024; Niko Partners forecasts China PC + mobile above US$57bn by 2027 and the total market above US$100bn annually by 2030. Growth is mid-single-digit and ARPU-driven; the domestic user base exceeds 600 million and is saturated.

INTERPRETATION: NetEase’s addressable market is no longer only domestic. The strategically important development of the last eighteen months is that its titles now compete for the global US$180bn+ games market on merit, on Steam and console, without a licence or distribution advantage — which means without the China discount attaching to the revenue.

Business Quality & Competitive Moat

Is the industry getting more or less competitive?

Less concentrated at the top and less rent-extractive in the middle. Content remains a Tencent/NetEase duopoly with a credible third-force challenger (miHoYo). What has genuinely changed is the distribution layer: the handset makers’ Hardcore Alliance held a ~50% take rate for a decade, and it is now being attacked simultaneously by the two publishers large enough to boycott and by the antitrust regulator (which forced Apple from 30% to 25% in March 2026). In Marathon’s capital-cycle language, capital and profit are migrating from distribution back to content.

Regulatory friction is also falling: 754 domestic approvals to 2026-05-26, up 19% year to date, with 2,300+ expected in 2026 — the most permissive posture since 2017.

How profitable is the business (ROIC, ROE)?

INTERPRETATION — and this is the key point: the reported 18.7% ROIC is cash-diluted and materially understates the business. Stripping the US$24.3bn of financial assets, operating invested capital is negative — accounts payable, accruals, taxes and RMB21.8bn of customer prepayments exceed the operating asset base. The games business consumes no shareholder capital at all, so return on operating capital is not computable in the ordinary way; the correct statement is that it is unbounded. Capital expenditure is 1.8% of revenue against D&A at 2.0%.

How profitable is the industry — how many competitors, what barriers to entry?

Barriers are unusually high and unusually specific: (i) the NPPA ISBN licence, without which no game may be commercially operated in China — a state-administered cap on supply; (ii) R&D scale — NetEase employs 25,382 people of whom 16,074 are in games and 47.8% are classified R&D, a fixed-cost base only two Chinese companies can carry; (iii) live-operations capability — running 100+ titles as continuous services with events, esports and balance patches. Industry profitability at the top is high (NetEase 31.8% operating margin; Tencent similar) and thin below it.

Can the business be easily understood?

Mostly. The revenue model (free-to-play with in-item monetisation, deferred on the balance sheet) is simple. Three things are not: the VIE structure, which requires understanding that shareholders hold a contractual rather than proprietary claim on 85% of revenue; the RMB/USD dual-currency structure, which injects large non-operating FX swings into GAAP earnings; and the US$3.07bn venture portfolio, whose marks and impairments make reported net income a poor guide to the business.

Can it be undermined by foreign low-cost labour?

No — the input is creative and engineering talent, and China is a low-cost source of it relative to Western studios. If anything the arbitrage runs in NetEase’s favour: it ships AAA-quality cross-platform product (Where Winds Meet on PC, mobile and console) at a cost structure Western publishers cannot match, which is one plausible reason its gross margin is 12–18 points above Nintendo’s and Square Enix’s.

Do brands matter?

At the franchise level, decisively — Fantasy Westward Journey, Onmyoji, Identity V, Naraka, and licensed-in Marvel and Minecraft are the assets. At the corporate level, barely — players choose Marvel Rivals, not NetEase. This asymmetry is why the Android boycott works for the big franchises and would not work for a new IP.

What is the nature of competition?

Competition for player time and wallet share, not for a contract or a shelf. It is won by content quality and live-operations cadence, and it resets with every major release. The competitive threat is not price; it is a better game.

Customers’ switching costs?

Real but non-contractual. A twenty-year Fantasy Westward Journey player has sunk character progression, purchased items, guild standing and social relationships that do not transfer. The financial evidence that these bind is the strongest available: a 22-year-old title setting a record 3.9m peak concurrent users, an 18-year-old MMO printing record quarterly revenue, and a games gross margin of 74.8% — none of which is achievable if the marginal player is free to leave and must be re-acquired.

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet?

Yes, and they are the most valuable assets the company owns. Internally developed game IP is expensed as R&D, not capitalised — RMB17.9bn of annualised R&D expense builds franchises carrying revenue for two decades and appears nowhere on the balance sheet. Goodwill is only US$265m (0.8% of assets) and disclosed intangibles US$980m, which are mostly licensed content, not owned IP. The stakes in listed subsidiaries are also carried at consolidation values rather than market: Youdao (NYSE: DAO) has a market capitalisation of ~US$2.0bn and NetEase Cloud Music (HKEX: 9899) ~HK$26.1bn (~US$3.3bn), against a total non-controlling-interest book of RMB4.7bn (~US$0.7bn).

Off-balance-sheet liabilities?

None disclosed as material. There are royalty, content and capital commitments disclosed in the 20-F commitments note; no pension obligation beyond statutory PRC contributions; no securitisation; no guarantee structure identified. The most significant “off-balance-sheet” item is conceptual rather than accounting: the VIE contractual arrangements themselves, which create the consolidation but not the ownership.

How conservative is the accounting?

Conservative on the operating side, mark-to-market on the investing side. Positives: game development expensed rather than capitalised; revenue deferred until delivered (contract liabilities RMB21.8bn); no aggressive capitalisation of software; capex at 0.91x D&A; a full valuation allowance carried against RMB3.95bn of deferred tax assets at certain entities. Negatives, or at least noise sources: the US$3.07bn long-term investment book produces annual impairments (RMB0.5bn / 1.3bn / 2.7bn in 2023/24/25, an accelerating series) and fair-value swings that flow through net income; and the non-GAAP measure NetEase publishes strips only share-based compensation, leaving the investment marks in.

How CapEx-hungry is the business?

Barely at all. FY2025 capital expenditure of US$293.5m on US$16.1bn of revenue is 1.8% of sales, against D&A of US$321.4m — a 0.91x replacement ratio on gross PP&E of US$2,354m. Applying the finite-life test used on asset-heavy operators (gross PP&E ÷ useful life versus actual capex), there is no undisclosed future call on cash. Combined with customer-funded working capital, this is about as capital-light as a US$16bn-revenue business gets.

Capital Allocation & Management

How much FCF does the business generate, how does management use it, what is the philosophy?

The philosophy has visibly changed, and not for the better. Five years ago NetEase returned three-quarters of free cash flow, weighted to buybacks. In FY2025 it returned 30%, weighted almost entirely to dividends, and cut repurchases 92.4% in a year when the stock spent the first half below $105 and free cash flow rose 36%. INTERPRETATION: the stated philosophy is a discretionary quarterly dividend “based upon our operations and earnings, cash flow, financial condition and other relevant factors”; the revealed philosophy is accumulation.

Significant acquisitions recently?

No — and this is a genuine strength. In each of the last five years NetEase spent less on acquisitions than it received from divestitures; it has been a net seller of businesses every year since 2021 (FY2025: US$38m out, US$435m in). Goodwill is 0.8% of assets. There is no serial-acquirer dilution and no impairment overhang — the direct contrast with the Take-Two/Zynga pattern (a US$9.7bn deal followed by ~US$5.9bn of goodwill impairment and ~60% dilution).

Buying back shares?

Nominally yes, effectively no. The current US$5.0bn authorisation commenced 2023-01-10 and was extended on 2025-11-20 by 36 months to 2029-01-09. As of 2026-03-31, 23.2 million ADSs had been repurchased for US$2.1bn — 42% of the authorisation in 3.2 years. FY2025 repurchases were US$91m, against US$521.6m of share-based compensation, so buybacks offset roughly 17% of the equity issued to employees.

Issuing large amounts of new shares to insiders?

Not large, but net-dilutive. Ordinary shares outstanding rose from 3,167,959,016 (Dec-2024) to 3,192,111,251 (Dec-2025), +0.76%, with Q1-2026 weighted diluted higher again. SBC is US$521.6m, 3.2% of revenue — modest by software/games standards (Roblox runs about 23% of revenue). A Second Amended and Restated 2019 Share Incentive Plan was filed 2026-07-15 and has not been analysed for pool expansion.

Compensation policy of directors/management?

FACT: aggregate 2025 cash compensation for all executive officers and directors in office was RMB44.5m (US$6.4m); RSU grants to executive officers would, if fully vested, represent less than 1% of outstanding ordinary shares. For a US$85bn company this is close to nominal.

INTERPRETATION: the low quantum is structural rather than virtuous — Ding owns 45.5% and is paid by the share price. As a foreign private issuer NetEase publishes no US-style compensation discussion and analysis with a performance-metric table, so the standard audit (does the incentive plan contain any per-share, ROIC, FCF or TSR metric?) cannot be run on public disclosure. That is a disclosure limitation and is recorded as such, not as a finding either way.

Motivations of management?

Ding is founder, CEO, director, 45.5% owner, sole director of the record-owner entity for his stake, settlor of the trust that holds it, and the personal registered equity holder of the principal VIEs. His economic interest is overwhelmingly the share price, which aligns him with minority holders — and there is no disclosed selling by him in five years of Rule 144 and Section 16 filings. The countervailing point is that no shareholder mechanism exists to change the distribution policy if he does not want it changed. The observable behaviour is that of an owner-operator building a franchise and hoarding a war chest, not an agent extracting rent.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer?

ADR — and the mechanics matter for the arithmetic. Each ADS represents five ordinary shares. This is the direct cause of a systematic vendor-data failure documented in the memo: ROIC.ai multiplies the ADS price by the ordinary-share count, inflating its reported enterprise value (US$417.7bn) and price-to-book (19.16x) by exactly 5x; yfinance compares an RMB revenue figure to a USD market capitalisation, producing a price-to-sales of 0.746. No vendor EV or P/B figure for NTES is usable. NetEase is also dual-primary listed on HKEX as 9999 (converted 2026-06-25), where the market capitalisation of HK$665.2bn reconciles exactly to the ADS-derived US$85.1bn. Not an MLP; no K-1.

Dividend policy?

Discretionary quarterly, declared per ordinary share and paid in US dollars. FY2025 declarations per ADS: Q1 $0.675, Q2 $0.570, Q3 $0.570, Q4 $1.160 = $2.975. Q1-2026: $0.720. Trailing yield 2.27% at $133.27, on a payout of ~38% of GAAP net income and ~28% of free cash flow. The dividend has grown from US$552m paid in FY2021 to US$1,977m in FY2025.

How profitable is the business?

TTM (Q2-2025 → Q1-2026): gross margin 65.7%, operating margin 33.3%, net margin 29.8%. Q1-2026 alone: gross 69.4%, operating 41.4%, net 34.9%. Games-segment gross margin 74.8% in Q1-2026. Five-year trajectory: gross margin +11.4 points and operating margin +12.1 points between FY2020 and FY2025, before the Q1-2026 step.

Is net income diverging from cash from operations?

Yes — favourably, and for an identifiable reason. FY2025 operating cash flow of RMB50.7bn (US$7,256m) was 1.50x net income of RMB34.8bn. The bridge is disclosed: non-cash add-backs of SBC RMB3.6bn, investment impairments RMB2.7bn, D&A RMB2.2bn, equity-method results RMB1.5bn and unrealised FX RMB0.7bn, plus a US$1,247m working-capital contribution dominated by the contract-liability build (short-term deferred revenue +41% year-on-year).

The correct treatment is to strip it: working-capital-neutral FCF is US$5,716m, not US$6,963m — a 6.7% yield on market capitalisation and 9.4% on enterprise value rather than 8.2% and 11.5%. Both are strong; the adjusted figure is the one to underwrite, because a deferred-revenue tailwind reverses when the portfolio stops growing.

Risks & Downside

What factors would cause the stock to decline?

In descending order of expected impact: (1) a PRC regulatory action on game monetisation — the 2023-12-22 draft spending rules removed 16.06% in one session on the heaviest volume of the five-year file; (2) reversal of the channel-fee gain, taking games gross margin back below 70%; (3) any concrete challenge to VIE consolidation; (4) a US listing or investment restriction (materially mitigated but not eliminated by the June 2026 dual-primary HK listing); (5) pipeline failure — Ananta slipping indefinitely, Sea of Elements missing its Q3-2026 window; (6) continued capital hoarding, which does not cause a decline so much as guarantee the discount persists; (7) non-operating noise — the Q4-2025 quarter showed a RMB1.67bn investment loss and RMB0.52bn FX loss cutting GAAP net income by a third with no operating deterioration whatever.

Risk of a catastrophic loss?

Low, and structurally so. Net cash is 28.5% of market capitalisation; there is no leverage; there is no single-asset or single-customer concentration; capital intensity is 1.8% of sales; and the business is funded by customer prepayments. No litigation is disclosed that management considers likely to be material. The operating business cannot realistically produce a catastrophic loss on any three-year view — a bear scenario in which the entire Q1-2026 margin gain reverses and revenue growth halves still supports roughly US$122 per ADS.

Chance of a total loss?

Very low, but not zero, and the route is legal rather than operational. The two paths are (i) a PRC action rendering the VIE contractual arrangements unenforceable, which would strip the consolidated claim on 85.2% of revenue, and (ii) a US action prohibiting ownership of, or delisting, the ADSs — the second of which is now materially mitigated because the ordinary shares are primary-listed in Hong Kong and holders have a fungible alternative. Neither is a business risk; both are jurisdictional. This is why the note argues about the correct size of the China discount rather than disputing that one is warranted.

Recent News & Events

Has the business environment changed recently?

Yes — three times in the last twelve months, all of them structural.

1. The Android channel revolt matured into margin. Since late 2024 NetEase has withdrawn its largest franchises from the OPPO, vivo, Xiaomi and Huawei app stores (OPPO shutdown reported 2025-11-05), bypassing a take rate of up to 50%. Games gross margin: 68.7% (FY2024) → 69.7% (FY2025) → 74.8% (Q1-2026).
2. Apple cut its China App Store commission from 30% to 25% effective 2026-03-15, under pressure from the State Administration for Market Regulation, saving Chinese developers a reported RMB6bn+ a year in aggregate. Present in only two weeks of the last reported quarter.
3. Licensing loosened. The NPPA approved 754 domestic and 25 imported titles to 2026-05-26, up 19% year to date, with 2,300+ expected in 2026 — the most permissive posture since 2017.

Significant acquisitions?

None. Net seller every year since 2021 (FY2025: US$38m of acquisitions against US$435m of divestiture proceeds).

Change in accounting policies?

None disclosed. The SEC’s Office of Technology did, however, run a comment-letter cycle on the FY2024 20-F (CORRESP 2025-07-31 and 2025-08-18; UPLOAD 2025-08-19) requiring disclosure changes: NetEase had to stop excluding Hong Kong and Macau from its definition of “China” for risk purposes, and had to restore risk language it had narrowed — specifically the statement that “The PRC government may also intervene with or influence our operations as it deems appropriate to further regulatory, political and societal goals at any time.” NetEase agreed and the sentence appears in the FY2025 20-F.

Recent changes — new markets, facilities, management?

• New markets: the first genuine Western live-service franchises. Marvel Rivals (40m registered players within two months of its December 2024 launch) and Where Winds Meet (15m → 80m+ cumulative players, 78% positive Steam rating, #2 on Steam’s global top-seller chart on its v1.6 update). Sword of Justice launched globally in Q4-2025.
• Domicile and listing: principal place of business moved from Hangzhou to Hong Kong (April 2026); dual-primary HKEX listing completed 2026-06-25 following a Migration Exchange Notice under Listing Rule 19C.11 (issued because more than 55% of worldwide trading by value had moved to Hong Kong). The stock rose +7.7% and +4.4% on the two sessions immediately after.
• Governance/reporting: the Holding Foreign Insiders Accountable Act (enacted 2025-12-18 in the FY2026 NDAA) extended Section 16(a) to foreign-private-issuer insiders effective 2026-03-18. Eight NetEase directors and officers filed initial Form 3s on 2026-03-17, and the first Form 4 in the company’s history was filed 2026-06-30 (General Counsel Paul Boltz Jr, 10,000 ADS at $128.30). A board-independence appointment was announced 2025-08-14. A Second Amended and Restated 2019 Share Incentive Plan was filed 2026-07-15.
• Management: no change to the CEO, CFO-function or key executives. Ding remains founder, CEO, director and 45.5% owner.

APPENDIX B — Source Appendix

NetEase, Inc. (NASDAQ: NTES / HKEX: 9999) — research date 2026-08-01

All sources accessed 2026-08-01 unless otherwise stated. Primary sources are listed first. Where a vendor figure was found to be erroneous it is listed with the error recorded, so that a future reader does not reuse it.

A. Primary — SEC filings (retrieved from SEC EDGAR, CIK 0001110646)

All of the above are publicly available from the SEC EDGAR full-text and filing-index systems under CIK 0001110646; 83 Exhibit 99 documents were retrieved from the corresponding 6-K accession directories.

B. Primary — management commentary

Caveat recorded: the ROIC transcript is machine-generated and substantially garbled on proper nouns (Sea of Elements → “Cement”; Where Winds Meet → “wellness”; Fantasy Westward Journey → “sentence virtual general”). Every game name and every figure used in the memo was cross-checked against the 20-F or the corresponding earnings release; no proper noun was taken from the transcript alone.

C. Quantitative data services

D. Industry, regulatory and news sources

E. Prior notes consulted for cross-read

The following prior notes by the same author were read for peer comparison, sector framing and China-ADR/VIE precedent. No conclusion in this note rests on them.

F. Analytical frameworks

• Bruce Greenwald & Judd Kahn, Competition Demystified — the barriers-to-entry taxonomy (customer captivity; economies of scale), the market-share-stability test, and the “which financial outcome would deteriorate?” moat test, applied to the competitive-position analysis.
• Edward Chancellor (ed.), Capital Returns (Marathon Asset Management) — supply-side capital-cycle analysis, applied to the migration of profit from the distribution layer back to content in the industry section.

G. Disclosed research limitations

1. No paid sell-side industry primers were consulted. Every claim in this note is sourced from public filings, company disclosure, public market data and published trade press.
2. No overseas revenue disclosure. NetEase does not publish a geographic revenue split. The single most important quantitative input to the “is the China discount still appropriate?” question is therefore unavailable, and the note says so rather than estimating it.
3. No US-style compensation disclosure. As a foreign private issuer NetEase files no DEF 14A with a CD&A performance-metric table, so the standard incentive-metric audit could not be run.
4. Blizzard licence term undisclosed. The FY2025 20-F gives the Minecraft licence expiry (2028) but not the term of the 2024 Blizzard arrangement now underpinning 9.5% of revenue.
5. The Apple commission benefit is directionally certain but not quantifiable from public disclosure, because NetEase reports no iOS/Android/PC/console revenue split. It is carried as an assumption, not a forecast.