Business Overview: Tencent Holdings Limited

HKEX: 0700 · 6 September 2026

Evidence base: Tencent's HKEXnews filings FY2015-FY2026 H1 (annual reports, interim reports and results announcements, IFRS, English) as the source of truth for the company, and independent industry, regulator and competitor sources for industry structure.

This is an analysis of the business. It is not a valuation and not a recommendation.

1. Executive Snapshot

2. What the Company Does

Tencent solves an unglamorous problem: a Chinese consumer needs one place to reach everyone they know. Weixin is that place, and it is free. The company earns nothing from the messaging itself. It earns from what a user does next inside the same app, and from the games it owns outright.

The unit: one Weixin monthly active user

The unit of economics is one Weixin/WeChat monthly active user. At 31 December 2025 there were 1,418m of them, up 2% on the year, against 697m ten years earlier (FY2015 and FY2025 annual reports). Group revenue of RMB751,766m divided across that base is RMB530 of revenue and RMB298 of gross profit per user for the year — figures computed from the filings, not reported by the company, which discloses no ARPU. Two caveats travel with them: the MAU count combines Mainland Weixin with overseas WeChat while 88% of revenue is Mainland, and International Games revenue of RMB77.4bn is earned outside the Weixin graph entirely. The per-user figure is therefore an approximation of the domestic engine rather than a clean average revenue per user.

Following one unit through to cash is short. The user installs Weixin free, links a bank card once, and thereafter tops up a wallet or buys virtual currency. That top-up is cash received before anything is earned: deferred revenue stood at RMB110,309m at the end of 2025, up from RMB100,097m a year earlier. Revenue is recognised as the user consumes items, watches an ad, or transacts. Working capital funds the business rather than consuming it, which is why operating cash flow of RMB303,052m in FY2025 exceeded reported profit of RMB224,842m.

Three revenue streams from one user base

Value Added Services was RMB369,281m in FY2025, 49% of revenue, at a 60% gross margin. Within it, games were RMB241,532m — 32% of group revenue — split between Domestic Games of RMB164.2bn (up 18%) and International Games of RMB77.4bn (up 33%, and above USD10bn for the first time). Social Networks contributed RMB127,749m from subscriptions, live streaming and music, growing 5%.

Marketing Services was RMB144,973m, 19% of revenue, at a 58% gross margin. This is advertising sold against Weixin's own surfaces — Moments, Video Accounts, Weixin Search and Mini Programs. Growth of 19% came, in the company's words, primarily from pricing and impressions, with only modest increases in ad load.

FinTech and Business Services was RMB229,435m, 31% of revenue, at a 51% gross margin. This is the toll booth: commercial payment fees, wealth-management and consumer-loan distribution, cloud services, and technology fees on merchant transactions inside Mini Shops. It grew 8% in FY2025, with Business Services growing at a high-teens rate and FinTech at a high single-digit rate.

What has been shed

Two lines have quietly gone. The FY2015 accounts carried an "Others" line of RMB4,726m that was largely e-commerce, a business Tencent exited in favour of holding equity in operators it does not run. And media advertising — inventory sold against Tencent Video and news properties — fell 8% to RMB14.3bn in FY2020 and no longer appears as a separate line: the current Marketing Services segment does not split social from media advertising at all. Both exits point the same way. Tencent has been retreating from advertising and commerce it must produce content or logistics to earn, toward advertising and fees it earns on traffic it already owns. The current "Others" segment, at RMB8,077m with a 4% gross margin in FY2025 and a negative gross margin in Q2 2026, is the residue.

3. Industry, Competitive Position and Moat

Tencent competes in four distinct industries that share one user base. Their structures are not alike, and Tencent's position in each is very different from the others.

Games: a licensed market where Tencent is the largest player

China's domestic games market produced RMB350.8bn of actual sales revenue in 2025, up 7.7%, and RMB188.5bn in the first half of 2026, up 12.2% (China Game Industry Report, GPC/CNG, December 2025 and July 2026). Tencent's Domestic Games revenue of RMB164.2bn is a large fraction of that pool, though the two figures are not directly comparable — one is recognised revenue under IFRS, the other an industry estimate of player spending. Globally the position is unambiguous: Tencent was the largest mobile game publisher by revenue in 2025 at USD8.75bn, with Honour of Kings alone above USD2bn (Sensor Tower, December 2025 and January 2026).

The barrier that actually binds is administrative. No game may monetise in China without an approval number from the National Press and Publication Administration. Approvals were frozen from March to 29 December 2018, and again from 22 July 2021 to 11 April 2022 — 263 days, with imported titles frozen longer still and resuming only in December 2022. Issuance has since normalised: 512 approvals in 2022, 1,075 in 2023, 1,416 in 2024, 1,771 in 2025 (the highest since 2018) and 1,362 through August 2026. This barrier cuts both ways. It caps how fast a challenger can build a portfolio, and it is the one mechanism capable of stopping Tencent's own pipeline without warning.

Two other regulatory events define the terrain. From 1 September 2021 under-18s may play online games only between 20:00 and 21:00 on Fridays, Saturdays, Sundays and public holidays — Tencent had disclosed that minors were 6.0% of its China game gross receipts in the fourth quarter of 2020, so the immediate revenue loss was small but the future cohort was cut off. And on 22 December 2023 the NPPA published draft measures capping in-game top-ups and banning login-reward inducements; Tencent's shares fell 12.7% that day (CNBC), the draft was pulled from the regulator's website on 23 January 2024, and no final version has been promulgated since. The lesson is not the rule, which never took effect, but the speed.

Advertising: third place, growing fastest

China's internet advertising market was RMB793.1bn in 2025 and grew 8.1% year on year in the second quarter of 2026 (QuestMobile, March and August 2026). Tencent is not the leader. On QuestMobile's measure of first-half 2025 spend, Taobao took 22.5%, Douyin 19.1% and WeChat 10.8%. But Tencent is compounding fastest of the large platforms: Marketing Services grew 19% in FY2025 and 22% in the second quarter of 2026, against Kuaishou's marketing revenue at +12.5% and +4.4% over the same periods and Alibaba's customer management revenue at +5% for its year to March 2026 (each company's own filings).

The stated mechanism is under-monetised inventory: Tencent says its ad load remains materially below peers', so price and impressions can both rise without degrading the feed. That claim cannot be checked. No independent measurement of Video Accounts' daily users, time spent or ad load exists, because QuestMobile measures applications rather than modules inside them, and Tencent discloses only that Video Accounts' total time spent grew more than 20%. The most important input to the fastest-growing profit line is unobservable to an outsider (unknown).

Cloud and payments: a distant third, and half a duopoly

In cloud Tencent is a follower. IDC put it at 8.8% of China's IaaS-plus-PaaS market in the first half of 2025, against Alibaba at 25.6%, Huawei 12.3%, China Telecom 11.4% and China Mobile 8.9%. For scale, Alibaba Cloud earned RMB158.1bn of revenue at roughly a 9% adjusted EBITA margin in its year to March 2026. Tencent discloses neither cloud revenue nor cloud margin; both sit inside a RMB229.4bn segment reported only down to gross profit (unknown).

In payments Tencent holds half of a duopoly whose economics are set by rule rather than competition. Alipay and Tenpay were 55.4% and 38.8% of third-party mobile payments in the first quarter of 2020, the last split published by a named data provider; Chinese research houses stopped disclosing company shares afterwards. The pool has stopped growing — the PBoC reports non-bank payment institutions handled 1,325bn transactions worth RMB337.8tn in 2025, with transaction count down about 0.7% and value up 1.9%. Since January 2019 all customer reserve funds sit at the central bank earning nothing, a rule codified in State Council Decree 768 effective 1 May 2024, and WeChat Pay's published standard merchant rate is 0.6% against a US all-card blended rate of 1.57% (Nilson Report, March 2025). Payments is a data-and-traffic business for Tencent, not a fee business; Tenpay was also fined RMB2.99bn by the PBoC in July 2023.

What a competitor cannot reproduce

No single Tencent product is unassailable. Douyin has more attention per user — 1,009m monthly users at over an hour and a half a day in March 2026 (QuestMobile) — and ByteDance has the better advertising stack. Alibaba has payments and merchants. NetEase makes excellent games. What none of them has is the combination: the address book, the payment credential and the merchant relationship living in the same application, so that a user who leaves loses all three simultaneously. That is not a product advantage that can be out-built; it is an accumulated position, and fourteen years of well-funded attempts have not dislodged it.

What could weaken it is not a better messenger. It is attention migrating to short video that Tencent must answer inside Weixin rather than win back; an agentic AI interface that makes the app layer itself less important, which is plainly why Tencent is testing Xiaowei inside Weixin; and a single administrative decision of the kind the 2021-22 freeze demonstrated. The kind of company that wins in these industries has licence-protected positions, owned intellectual property, and enough internal cash to fund a capital cycle without asking anyone. Tencent is that kind of company.

4. Growth Engine

Revenue grew 14% to RMB751.8bn in FY2025 and 10% to RMB401.2bn in the first half of 2026. Tencent publishes no organic growth rate (unknown), but the decomposition can be reconstructed. Cash paid for business combinations was RMB11,541m in FY2025 — 1.5% of revenue — and no material consolidation occurred, so FY2025 growth was substantially organic (inferred). That has not always been true: FY2020's 28% growth carried the full-year effect of consolidating Supercell and the consolidation of HUYA's live-broadcast revenue, both named in that year's annual report. The one adjustment Tencent does disclose is currency on International Games: +33% reported against +32% constant currency in FY2025, then -0.8% reported against +4% constant currency in the second quarter of 2026. The entire reported decline in that quarter was the exchange rate.

Ranked by contribution to growth, most to least:

• Advertising price, not advertising volume — structural. Marketing Services grew 19% in FY2025 and 22% in Q2 2026, driven by pricing (AI-driven targeting, advertisers generating more creatives, and a rising share of closed-loop ads that click through to Mini Shops and Mini Games) with only modest increases in ad load. Because the yield is rising on inventory Tencent already owns, incremental gross margin is high and the line grows roughly three times faster than the market.
• Domestic games gross receipts from evergreen titles — structural, with a cyclical release overlay. Up 18% in FY2025 and 17% in Q2 2026. Delta Force is the first new domestic title in several years to reach evergreen scale (defined by the company as over 5m average quarterly mobile DAU and over RMB4bn of annual gross receipts), joining Honour of Kings and Peacekeeper Elite. The base is structural; the timing of any single new title is not.
• International Games — management-driven. RMB77.4bn in FY2025 from Supercell, PUBG MOBILE, Wuthering Waves and Miniclip. This is the only line where Tencent competes with no regulatory shelter and no home social graph, which makes it the cleanest read on whether the company can make games people choose rather than games people find.
• Cloud and merchant technology fees inside FinTech and Business Services — structural but sub-scale. Business Services grew at a high-teens rate in FY2025 while FinTech grew high single digit, producing a segment total of 8%. Growth here is real, but the segment is dominated by its slower half, and the disclosure does not let an outsider size the faster one.
• Social Networks — no longer a driver. Up 5% in FY2025 and 0.8% in Q2 2026, with fee-based subscriptions falling to 259m from 264m a year earlier. Subscription video and music are mature.

What is conspicuously absent from the growth ledger is artificial intelligence. Yuanbao, WorkBuddy, CodeBuddy and Xiaowei are described in the filings in units of users, retention and token consumption — never revenue (unknown). AI currently appears in Tencent's accounts as a cost and a capital commitment, and as an efficiency gain inside advertising, not as a business line.

5. Margin, Cash and Capital Allocation

The gross margin U-curve, and why it happened twice

Group gross margin fell from 59.5% in 2015 to 44.4% by 2019, bottomed at 43.1% in 2022, and recovered to 56.2% in 2025. Both halves have the same explanation running in opposite directions. The decline was the deliberate construction of payments and cloud — businesses that carry bank channel fees, settlement costs, bandwidth and server depreciation inside cost of revenue. FinTech and Business Services ran a 28% gross margin in FY2020. The recovery is those same businesses reaching scale while the high-margin lines grew faster: FinTech and Business Services went from 28% in FY2020 to 47% in FY2024 and 51% in FY2025, and within advertising the fastest-growing surfaces (Video Accounts and Weixin Search) are the cheapest to serve. Margin here is a mix outcome, not a pricing outcome, which is why it moved so far in both directions.

Not comparable across all four columns. Segment definitions changed twice: FY2015 reported VAS, online advertising and Others; FinTech and Business Services was created in FY2019; and "Online Advertising" was renamed Marketing Services from FY2023. FY2015's 59.5% gross margin describes a company that was almost entirely games and social, not the same mix at a different margin. Revenue and gross profit per MAU are computed, not reported.

Where the cash went

FY2025 capital allocation, ranked by cash out: capital expenditure of RMB79,198m; share repurchases of RMB73,312m, being 153.4m shares cancelled for roughly HKD80.0bn; dividends of RMB37,535m; and RMB24,637m into associates plus RMB11,541m on business combinations. The company returned RMB110.8bn to shareholders and spent RMB79.2bn on capacity in the same year, entirely from RMB303.1bn of operating cash flow, and still ended with net cash of RMB107.1bn. The dividend has compounded steadily rather than dramatically — a final HKD5.30 per share for FY2025, from HKD4.50 for FY2024, HKD1.60 for FY2020 and HKD0.47 for FY2015. Buybacks reduced the share count from 9,224.9m to 9,120.2m, a net 1.1% after option exercises and share awards, and were smaller than FY2024's 307.2m shares for roughly HKD112.0bn.

The behaviour this reveals is a management that treats buybacks as the flexible line and capital expenditure as the committed one. That ranking was tested in 2026 and it inverted immediately.

What happened after the reported year

These figures post-date the FY2025 annual report and come from the interim results announcement for the six months to 30 June 2026. Capital expenditure was RMB84,720m in the first half of 2026 — more than the whole of FY2025 — and RMB52,784m in the second quarter alone, up 176% year on year. Buybacks fell to roughly HKD24.4bn for the half. Net cash fell from RMB146,860m at 31 March 2026 to RMB58,191m at 30 June, after RMB59.3bn of capital expenditure payments and RMB41.6bn of dividends. Free cash flow was negative RMB13.8bn in the quarter; the company states it would have been positive RMB37.6bn excluding prepayments for compute procurement.

The investment portfolio is a second, separate engine and it runs on its own cycle. Listed holdings were worth RMB672.7bn at the end of 2025, up from RMB569.8bn, with unlisted holdings carried at RMB363.1bn; by 30 June 2026 listed holdings had fallen to RMB487.2bn while unlisted rose to RMB387.9bn. The portfolio contributed RMB29.7bn of return in FY2025. It also explains the shape of the most recent quarter: share of associates' and joint ventures' results swung from a profit of RMB4.5bn to a loss of RMB10.0bn, which is why reported profit attributable to shareholders was flat at +0.7% in Q2 2026 while non-IFRS profit, which strips these effects out, rose 9%.

6. Cyclicality, Constraints and What to Monitor

Tencent's end markets are overwhelmingly domestic: RMB662.1bn of FY2025 revenue, 88%, came from Chinese Mainland. Within that, advertising is the cyclical line, because marketing budgets are discretionary and cut first. Payment fee income tracks retail consumption almost mechanically. Games spending has been the defensive line: small recurring outlays inside an entertainment habit, which held up through both the 2022 downturn and the regulatory shock that accompanied it.

Where the business sits in its own cycle

On demand, at or near a peak: FY2025 revenue, gross profit and gross margin were all records, and 56.2% is the highest gross margin since 2016. On capital, at the top of a spending cycle that has no precedent in this ten-year archive. Capital expenditure ran RMB32.4bn for all of FY2020, RMB79.2bn for FY2025, and RMB84.7bn in six months of 2026. That step-up turned quarterly free cash flow negative for the first time in the period covered here. Gross margin has not yet felt it — 58% in Q2 2026, still up year on year — because much of the spend sits in prepayments and in assets not yet depreciating at full run-rate. The margin peak and the capital peak are therefore not the same moment, and the second one has not yet passed through the income statement.

Constraints that bind

• Game approval throughput. Every monetised title needs an NPPA number, and issuance has twice been suspended for months at a time.
• Payment float. All customer reserve funds have sat at the PBoC earning nothing since January 2019, removing the interest income that made payment processing profitable elsewhere in the world.
• Consumer credit economics. Under the CBIRC's February 2021 rules the bank partner must fund at least 30% of each joint loan and bears credit risk on its own share, which caps what Tencent can capture through its 30% stake in WeBank. WeBank itself is going backwards: FY2025 revenue fell 4.8% and loans fell 3.4%, with net profit of RMB11.0bn up just 1.0% on assets of RMB766.3bn.
• Compute access. US export rules moved H200 shipments to case-by-case review in January 2026, after earlier rules blocked the A800 and H800 in October 2023. A capital plan funded in renminbi can be interrupted by a rule written elsewhere.

The downside case, as a mechanism

The damaging combination is not a games ban. It is an advertising recession arriving while the 2026 capital expenditure begins to depreciate. Marketing Services carries a 57-58% gross margin, so a decline there drops to profit with almost no cushion, and unlike capital expenditure it cannot be deferred. In that sequence the mix effect that produced the 2024-25 margin recovery runs in reverse — high-margin revenue shrinks while a much larger depreciation charge lands in cost of revenue — and the gross margin retraces without anything happening to the competitive position at all.

7. Risks, Unknowns and Questions for Deeper Work

The risks below are those that cyclicality does not capture. They are ordered by how badly they compound.

• The operating company is controlled by contract, not owned. PRC rules bar foreign ownership of value-added telecom licences, so Tencent Computer — which holds the licences the business runs on — is legally owned by PRC-citizen founders and controlled through Structure Contracts. Every claim in this document about Tencent's revenue assumes those contracts continue to be honoured and enforced.
• One administrative decision can halt the largest profit line. Domestic Games is 22% of revenue and, at a 60% segment gross margin against a 56% group margin, more than that of gross profit. The 2021-22 approval freeze ran 263 days. Nothing about the current normalised issuance rate is contractual.
• The AI capital cycle has no revenue attached to it in the disclosure. Capital expenditure rose 176% year on year in Q2 2026 while every AI product is described in usage terms only. If depreciation arrives on schedule and revenue does not, the margin gains of 2024-25 reverse mechanically rather than competitively — and this compounds with the advertising-recession mechanism above, because the same margin absorbs both.
• The financial holding company ordered in April 2021 has still not been licensed. As of April 2026 only three such licences existed in China and Tencent held none. Five years of unresolved regulatory status sits over payments, WeBank and wealth-management distribution, with no published timetable.
• Two shareholders hold 31.6% between them. Prosus/MIH holds 22.80% and has sold in the past; Ma Huateng's vehicle holds 8.82%. Neither is a risk to the business, but both shape what the register can absorb.

What the sources could not answer

• No segment operating profit. Tencent's chief operating decision-makers assess segments on revenue and gross profit only; selling, general, administrative and R&D costs are managed centrally and not allocated. It is impossible to say from the filings whether cloud or payments earns an operating return at all.
• No cloud revenue or margin, and no AI revenue, disclosed anywhere.
• No Video Accounts daily users, time spent per user or ad load — from Tencent or from any independent source. The input that matters most to the fastest-growing line is unmeasurable outside the company.
• No credible Alipay/Tenpay market-share split after the first quarter of 2020. Chinese data providers stopped publishing company shares.
• No first-party transcripts. Tencent does not publish them, so prepared remarks are recoverable from results announcements but management's unscripted answers are not available from an official source.
• No Licaitong assets under management, and no disclosure of what share of FinTech revenue comes from payments versus wealth management versus lending.

Questions before forming a view

• What operating margin does Business Services actually earn at RMB229bn of segment revenue, once centrally-held costs are allocated? Without this, the case for cloud rests on a gross margin that excludes most of its cost base.
• How much of the 2026 capital expenditure serves internal model training versus paying external cloud demand? The first is a bet; the second is a fulfilled order book.
• What happens to Marketing Services growth once the AI targeting uplift is fully in the base — does 20%+ growth persist against an 8% market, or was this a step change now largely taken?

8. Investor Takeaways

• Tencent is an attention-and-settlement utility, not a games company that also does other things. Games are 32% of revenue; the durable asset is the 1,418m-user Weixin graph that makes the other 68% possible.
• The economic engine is a prepaid, capital-light toll on one user base — RMB530 of revenue and RMB298 of gross profit per Weixin user in FY2025, funded by customers who pay before they consume.
• The main growth lever is advertising price on inventory Tencent already owns, running at 19-22% against a market growing 8%, with no ad-load increase doing the work.
• What could break the story is not competition. It is a regulatory stop of the kind already demonstrated twice, or an AI capital cycle whose depreciation arrives before its revenue does.
• Monitor capital expenditure against free cash flow, Marketing Services growth against the QuestMobile market rate, and the monthly NPPA approval list. Everything else is detail.