01 — Executive summaryA quality franchise priced as a value trap
Tencent reported revenue of RMB 204.8 billion for the three months ended 30 June 2026, up 11% year on year, and non-IFRS profit attributable to shareholders of RMB 68.4 billion, up 9%. The reported IFRS figure was RMB 56.0 billion, up 0.7%. Both are correct, and the gap between them is the investment case in miniature.
This report applies a value-investing framework, not an AI narrative. On that method Tencent has compounded revenue at 10% a year, expanded gross margin fourteen points in five years, earned a return on equity of twice its cost of capital and converted profit into cash at 1.35 times — at a multiple a value investor associates with terminal decline.
The reason is not the past but the present. Capital expenditure rose 176% to RMB 52.8 billion in the quarter, first-half capex exceeds the whole of FY2025, free cash flow was negative RMB 13.8 billion, and the buyback has been cut roughly 45% to help pay for it. Whether that is value creation is unresolved.
What can be resolved is the price. Strip out RMB 58.2 billion of net cash and an RMB 875.1 billion investment book marked at three-quarters of book, and the operating business trades at 9.9 times forward non-IFRS earnings. Our base case values the shares at HK$565, a 24.7% margin of safety inside the 20–50% band this framework requires. We rate the shares Constructive with a HK$565 target.
02 — The macro backdropA slower China, and what it does to a platform
Tencent is a leveraged play on Chinese consumption: advertising budgets and payment volumes are functions of nominal spending. A top-down read starts with the growth rate China is running at.
The economy has settled into a 4–5% band. Real GDP grew 5.0% in 2024 and an estimated 4.8% in 2025, and the IMF lifted its 2026 forecast to 4.6%. Nominal retail sales, the series that matters more for a platform, have run near 4% after a deflationary 2022–2024 stretch.
This cuts both ways. A platform cannot outgrow nominal spending forever, so 10% revenue growth must eventually decelerate. But Tencent's growth comes from taking share of a fixed advertising pool and monetising existing users more efficiently, not from a rising tide — a lower-quality backdrop than the 2010s, and the reason the shares trade where they do. Policy has normalised since 2023.
03 — Industry structureAn oligopoly with high walls and a state in the room
Chinese consumer internet passes the first two of the framework's industry tests — growth and barriers — and is ambiguous on the third, competition.
The industry is mature but not shrinking: growth comes from monetisation per user rather than new users, which rewards scale over novelty. The barriers are formidable. A messaging network with 1.44 billion monthly users cannot be replicated by capital — the value is the graph, and the graph took twenty years to build. Payment licences, game approvals and the mini-program ecosystem cannot be bought.
Competition is not the comfortable duopoly a value investor would prefer. ByteDance competes for attention and advertising, Alibaba for cloud and commerce, and all three now spend heavily on AI. The distinguishing feature is that a fourth party — the regulator — sits inside the oligopoly, able to change the terms of monetisation unilaterally. That belongs in the discount rate.
04 — The business modelThree engines on one chassis
Tencent is usually called a conglomerate, which obscures more than it explains. It is three businesses of very different quality sharing one distribution channel and one balance sheet.
Value-added services is games plus social networks: RMB 98.4 billion, 48% of revenue, at a 60% gross margin, hit-driven but durable because the largest titles are over a decade old and behave like annuities. Marketing services is advertising across Weixin Moments, Video Accounts and Mini Games: RMB 43.6 billion, the smallest line and the fastest growing at 22%, and the best incremental revenue in the company: an impression sold carries almost no marginal cost. FinTech and business services is payments, wealth management, lending and cloud: RMB 60.3 billion at a 51% gross margin, a bank and a utility stitched together.
All three engines are supplied by the same distribution asset. Tencent pays nothing to acquire the audience that plays its games or clicks its advertisements, which is why the consolidated margin exceeds what the parts alone suggest.
05 — Revenue qualityGrowth is real, but it is not evenly distributed
The composition of growth matters more than its level. Revenue rose 11%, but that average conceals a 22% advertising line, a flat social-networks line, domestic games up 17% and an international games business that went backwards.
Two things follow. The fastest-growing revenue is the highest-margin revenue, so a quarter in which ads grow 22% mechanically lifts group margin. And international games, Tencent's best story in 2025 at plus 33%, stalled at minus 0.8% as reported, or plus 4% in constant currency — a currency effect, not a demand failure, but growth has rotated back to a tighter home market.
Growth at Tencent comes mostly from share gains and pricing: advertising growth comes from pricing rather than ad load, and the games line from titles owned outright. Neither depends on the Chinese consumer accelerating.
06 — The moatThe one asset a competitor cannot buy
Tencent's moat is a network: Weixin becomes more valuable to each user as more users, merchants and mini programs join it, and the cost of leaving is every conversation, contact and payment record the user depends on — a switching cost measured in years of accumulated behaviour.
Weixin reached 1,439 million combined monthly active users at 30 June 2026, up 2% year on year, and QQ 520 million, down 2%. The user base is no longer a growth story and does not need to be: the moat is the installed base and everything built on top of it. More than four million mini programs and a payments rail make Weixin closer to an operating system for Chinese daily life.
The counterweight is that a moat no one can buy can still be routed around. ByteDance built an attention franchise without a social graph. The moat protects communication, payments and mini programs; it does not guarantee that the next form of commerce is born inside Weixin.
07 — Pricing powerThe advertising line is the proof
A moat is only worth what it lets the owner charge. The cleanest evidence is marketing services, where revenue has grown at roughly a fifth a year without any material increase in ad load.
Advertising grew 20% in FY2024, 19% in FY2025 and 22% in the June quarter, and management attributes the increase to pricing: a better AI recommendation model, the AIM+ automated campaign tool, and closed-loop formats that let Mini Shops and Mini Games advertise against a measurable transaction.
Tencent has run a deliberately low ad load for a decade while competitors monetise far more impressions per user, and that restraint is now an option: it can raise revenue by selling the same audience more effectively, without adding inventory, at close to zero marginal cost. A deceleration into single digits would be the earliest warning that the moat is being tested.
08 — Management & allocationTwenty years of buying back stock, then a change of mind
Capital allocation is the clearest window into management quality.
Tencent's record is among the better ones in Asia: a growing dividend paired with an aggressive buyback — HKD 80.0 billion in FY2025, executed at prices that in hindsight look low. In 2026 that changed. Through 9 October the company had repurchased HKD 32.0 billion, an annualised pace roughly 45% below last year, with the capital redirected into compute. The dividend was not cut: the HKD 5.30 final dividend, up 18%, was paid on 1 June 2026.
Two readings are possible and both should be held. The generous one is that management sees a higher marginal return in AI infrastructure than in its own stock. The sceptical one is that a company buying back stock at twelve times earnings was creating certain per-share value, while one building data centres makes a bet whose return will not be visible for years. We lean to the first, but the buyback did continue through October at around HK$425.
On governance the structure is standard for a Hong Kong-listed Chinese technology company: a founder-led board, dual-class voting rights, and no history of accounting restatement or sanction that would trigger the framework's integrity screen. Concentrated voting power is a real cost to minority holders and is why we discount the peer multiple; it is not, on the record, a reason to avoid the equity.
09 — ProfitabilityA return on equity a utility would not recognise
The framework's core profitability test is a sustained, high return on equity, decomposed to show where it comes from. Tencent passes it with room to spare.
On closing equity attributable to shareholders, the FY2025 return was 19.5% on an IFRS basis and 22.5% on a non-IFRS basis; on average equity, 21.1% and 24.4%. Both are roughly twice any reasonable cost of equity. The IFRS figure is dragged by non-cash investee marks through the associate line, which is why we read the non-IFRS number as the operating truth.
A DuPont decomposition shows how the return is built: net margin 29.9%, asset turnover 0.39 times, equity multiplier 1.79 times. It is a margin story, not a leverage story — the opposite of the profile the framework warns against — and the multiplier is low precisely because the company holds net cash. The qualification is that book equity is depressed by a decade of buybacks and by a fair-value reserve that swings with Chinese equity marks, so the ratio flatters in years when marks rise.
10 — Margin expansionFourteen points in five years
The gross margin is the chart to keep. Consolidated gross margin was 43.9% in 2021 and 57.8% in the June 2026 quarter. Fourteen points on a revenue base approaching RMB 800 billion is not an accounting effect; it is a structural change in what Tencent sells.
Three shifts produced it. Internally developed games displaced licensed content, and internal titles carry no revenue share. Video Accounts and Weixin Search advertising replaced low-margin resale lines. And cloud and payments, both marginal in 2021, grew into their fixed cost.
The consequence for valuation is direct: earnings grow faster than revenue even with no further margin expansion, because the mix keeps improving. Non-IFRS operating profit excluding new AI products grew 19% in the June quarter against 11% revenue growth — operating leverage, and the thing most at risk if advertising slows.
11 — Cash conversionProfit that arrives as cash, not as receivables
Reported profit is an opinion; cash is a fact. The framework's second quantitative test is whether operating cash flow corroborates the income statement: a conversion ratio above one separates real earnings from accounting ones.
Tencent clears the bar. Operating cash flow was RMB 258.5 billion against reported profit of RMB 194.1 billion in FY2024, and RMB 303.1 billion against RMB 224.8 billion in FY2025 — conversion ratios of 1.33 and 1.35 times. The excess over profit reflects heavy depreciation and deferred revenue.
The caveat is specific to 2026. June-quarter operating cash flow of RMB 52.7 billion is reduced by large prepayments for compute — cash paid today for capacity delivered later. The conversion ratio will therefore look worse before it looks better, which is why the next section matters more than the cash flow statement alone.
12 — Capital intensityThe quarter in which the sign flipped
The most important change at Tencent in 2026 is not in the income statement.
Capital expenditure was RMB 19.1 billion in the June 2025 quarter and RMB 52.8 billion a year later, up 176%. First-half 2026 capex of RMB 84.7 billion exceeded the RMB 79.2 billion spent in all of FY2025. Management notes that spending is constrained by when advanced accelerators can be procured, which makes quarterly capex a poor guide to intent; the direction is unambiguous.
The consequence showed up in the cash flow statement. Operating cash flow of RMB 52.7 billion was absorbed by RMB 59.3 billion of capital expenditure, RMB 5.0 billion of media content and RMB 2.2 billion of lease principal. Free cash flow was negative RMB 13.8 billion — the first time in memory the company has consumed cash in a quarter while reporting record revenue. Tencent's qualification is fair: excluding compute prepayments, free cash flow would have been positive RMB 37.6 billion. Both are true, and the gap between them is the central accounting question in this report.
This is the crux, and a value investor should be uncomfortable with it. The framework prefers capital-light businesses because they convert profit into distributable cash with high certainty, and Tencent is giving some of that up. If capex holds at the second-quarter run rate, the annual figure approaches RMB 210 billion, which operating cash flow cannot fund alongside the dividend and the repurchase. The bear case rests on exactly that arithmetic.
13 — The balance sheetSolvency is not the problem; the buffer is
The framework screens for leverage first, on the view that a value investor's first job is to survive. Tencent passes easily, which is worth stating plainly because "net cash fell" can be read as deterioration when it is not.
Total liabilities were 43.4% of total assets at 30 June 2026, up from 39.1% at the end of 2025 but modest for a company of this scale. The current ratio was 1.29 times, inventory is negligible, and operating profit covers interest expense more than twenty times. This is not a company funding a capital programme with borrowed money.
What is falling is the cash buffer. Net cash — cash, term deposits and liquid treasury products less borrowings and notes payable — fell from RMB 107.1 billion at the end of 2025 to RMB 58.2 billion at 30 June 2026. The company is funding the build from its own balance sheet rather than debt, which is conservative, but the cushion that would absorb a bad year is thinner. This is a capital-allocation question, not a solvency one.
14 — The investment bookA second balance sheet the income statement hides
Two decades of minority stakes in Chinese internet companies have left Tencent owning a portfolio roughly a quarter the size of its own market capitalisation. For a value investor this is the most important asset the earnings multiple cannot see.
At 30 June 2026 the group held RMB 487.2 billion of listed investees at fair value and RMB 387.9 billion of unlisted investees at carrying value — RMB 875.1 billion in total. The listed book fell 28% from RMB 672.7 billion at the end of 2025; the unlisted book rose from RMB 363.1 billion.
That asymmetry matters. The listed half is marked to market every quarter and is volatile and transparent; the unlisted half is carried at cost or the last funding round and is neither. It also explains the reported-earnings confusion: a portfolio this size will keep producing several-billion-renminbi swings through the associate line, in both directions, unrelated to games, advertising or payments.
We value the portfolio at 75% of book in our base case, a discount for tax on eventual gains and the illiquidity of the unlisted half. That is HK$79.4 per share, 19% of the current price — a genuine margin of safety, not an accounting artefact: the listed half could be liquidated, and the unlisted half is real businesses with real revenues.
15 — Intrinsic valueA discounted cash flow that brackets the price
The framework's preferred method is a discounted cash flow on sustainable free cash flow, because it forces every assumption into the open. We present it as a range rather than a point, because a single output invites false precision.
The starting figure is normalised owner earnings of RMB 235 billion for 2026: the non-IFRS profit run rate less the excess of capital expenditure over depreciation. We grow that at 8% for five years and 3.5% for the next five, with terminal growth of 2–3%. Across a cost of capital of 9–11%, the operating business is worth HK$430 to HK$614 a share; the centre gives HK$504, and every cell clears the HK$425.20 price.
The DCF covers only the operating business; the investment portfolio and net cash are added in the sum-of-the-parts that follows. On central assumptions, intrinsic value including those stubs is roughly HK$590. We treat this as a cross-check rather than a target, because it rests on a terminal growth assumption no one can verify. A DCF that landed far above the market price on generous assumptions would be a warning, not a conclusion; this one brackets the price from a conservative starting point.
16 — Relative valuationCheap against its history and against its peers
The framework also asks how the shares are priced against its own history and against comparable businesses.
Against its own history, Tencent is at the low end of where it has traded: roughly 12.4 times our FY2026 non-IFRS earnings estimate, against one-, three- and five-year medians of about 16, 18 and 20 times. The medians are indicative, but the direction is robust.
Against peers the discount is larger still: roughly half the forward multiple of the US platforms and below Alibaba. That gap is not a growth discount — Tencent grew revenue 11% last quarter, in the same band as its peers — and not a balance-sheet discount, since Tencent carries net cash. It is a jurisdiction discount. It is wider than the fundamentals justify, but the correct response is to size the position for the risk rather than assume it closes.
On the remaining measures: price-to-book is 3.1 times, which looks expensive until you note that book equity is depressed by buybacks and the fair-value reserve; price-to-sales is 4.2 times on FY2026E revenue; and PEG is roughly 1.2 on a 10% growth assumption. With an ROE of twice the cost of equity and cash conversion above 1.3 times, the picture is a high-quality business priced as though the quality is temporary.
17 — Margin of safetyTwo-fifths of the price is not the operating business
The framework buys only where the price sits 20–50% below calculated intrinsic value. That requires a valuation the investor can defend, and for Tencent the most defensible method is a sum-of-the-parts, because the company is three assets stacked in one security.
At HK$425.20, net cash of RMB 58.2 billion is HK$7.0 per share. The investment book of RMB 875.1 billion, marked at 75%, is HK$79.4. What is left — HK$338.8 — is the operating business, which against FY2026 non-IFRS earnings of RMB 31.13 per share is 9.9 times.
Our base case applies 14 times to those earnings — still a discount to the global platform peers — and adds the same two stubs back, giving HK$565. That is 24.7% above the market price, inside the framework's required band and the level at which we set our target. The bull case gives HK$684 and the bear case HK$368; the weighted value is HK$546.
The composition of the price is the point. Two-fifths of what an investor pays is not the operating business at all: it is a portfolio of Chinese internet assets and a net cash balance. A buyer at this price pays under ten times earnings for the operating franchise and receives the rest as an attached asset — a margin of safety in the framework's terms, created because the market applies one discounted multiple to a company that is not one thing.
18 — ScenariosWhat the price already assumes
We run three cases, each built from two explicit inputs: the multiple applied to forward non-IFRS operating earnings, and the discount applied to the investment book. No third assumption is hidden inside.
| Scenario | Operating multiple | Investment book | Value per share | Weight |
|---|---|---|---|---|
| Bear | 9.0x | 50% of book | HK$368 | 25% |
| Base | 14.0x | 75% of book | HK$565 | 50% |
| Bull | 17.0x | 90% of book | HK$684 | 25% |
Operating multiple applied to FY2026E non-IFRS earnings of RMB 31.13 per share; net cash of HK$7.0 per share added in every case. Probability-weighted value HK$546. Farstar estimates.
The distribution is the point. Our bear case, HK$368, is 13% below the current price and assumes the operating business never re-rates from where it trades today and the investment portfolio is worth half its book value. Our bull case, HK$684, applies a multiple still below where the global platform peers trade. The asymmetry is roughly one to three in favour of the upside, and the reason is not optimism about AI — the starting multiple already assumes the AI investment earns nothing. The bear case is not a tail; it is close to the price.
19 — The bull caseIn one page
The moat is a network, not a product. Weixin carries 1.44 billion monthly users, more than four million mini programs and a payments rail — an asset capital cannot replicate, and why the franchise has earned twice its cost of equity.
The core compounds faster than reported. Gross margin has risen fourteen points in five years and non-IFRS operating profit excluding AI products grew 19% against 11% revenue growth. Reported profit is depressed by a non-cash associate line, not by the operating business.
The accounts are clean. Operating cash flow has exceeded reported profit by more than a third in each of the last two years, leverage is modest, and the company carries net cash.
Two-fifths of the price is not the business. Strip out net cash and the investment book and the operating franchise trades at 9.9 times forward non-IFRS earnings. A re-rating to 17 times, still below global peers, gives HK$684.
20 — The bear caseIn one page
The capital is spent and the return is unproven. Capex tripled, free cash flow went negative, the buyback was halved, and the AI products cost RMB 10.5 billion in the quarter without generating any disclosed revenue.
Capital intensity may be permanent, not temporary. If capex holds near the second-quarter run rate, the annual figure approaches RMB 210 billion, which operating cash flow cannot fund alongside the dividend and the repurchase. A capital-light franchise is worth more than a capital-heavy one at the same earnings.
The moat may not extend to the next platform. ByteDance built a dominant attention franchise without a social graph. If the next form of commerce is born outside Weixin, the network that protects messaging and payments does not automatically protect the future.
The discount may be deserved. Game monetisation rules, platform antitrust, AI-content regulation and export controls on accelerators are all live, and the regulator is a participant in this oligopoly rather than an observer. A lower multiple than a US platform may be the correct price of jurisdiction risk.
21 — Risk screeningWhat could break the thesis
The framework screens fundamental, financial, external and governance risks before a position is built; the table states each and how it would show up.
| Risk | Type | Probability | Impact | How it would show up |
|---|---|---|---|---|
| Capex stays above RMB 50bn a quarter into 2027 | Financial | Medium | High | Net cash approaches zero; buyback suspended; dividend growth stops |
| Advertising growth falls to single digits | Fundamental | Medium | High | Group gross margin stops expanding; the 19% ex-AI growth breaks |
| The AI products never earn their cost of capital | Fundamental | Medium | High | No disclosed AI revenue by 2027; capex written down or slowed abruptly |
| Investment portfolio marks fall again | Financial | High | Medium | IFRS profit falls sharply while non-IFRS is unchanged; sentiment follows the headline |
| Chinese consumer confidence weakens further | External | Medium | Medium | Payments, ad spend and mini-shop GMV decelerate together |
| Accelerator supply constraint worsens | External | Medium | Medium | Capex undershoots guidance; model training delayed; cloud growth stalls |
| Games approval or monetisation restriction | External | Low | High | Domestic games revenue growth turns negative in a single quarter |
| Regulatory action on AI content or platform data | External | Low | High | Agent rollout delayed; compliance costs; limits on Weixin data use |
| Governance: voting power stays concentrated | Governance | High | Low | Minority holders cannot influence capital allocation; discount persists |
| Geopolitical escalation affecting HK listings | External | Low | High | Multiple compression independent of earnings; fund-flow effects |
| Key-person dependence on the founder group | Governance | Low | Medium | Strategy discontinuity; succession uncertainty |
| Dividend or buyback cut to fund capex | Financial | Medium | Medium | Capital return falls for a second year; the equity becomes a pure growth bet |
22 — The holding decisionHow we would own it, and what would change our mind
The framework is explicit: buy in batches at low valuations, hold while the fundamentals hold, and sell when the price exceeds intrinsic value or the logic deteriorates.
Buying criteria — met
A 24.7% margin of safety; a moat that has produced a return on equity of twice the cost of capital; operating cash flow above reported profit by more than a third; net cash; and no history of restatement or sanction. Five of the framework's six buy conditions are met. The sixth — visibility on the return of the current capital programme — keeps this constructive rather than aggressive.
Position construction — in batches
We would build in thirds: a first tranche now, a second on evidence that free cash flow has turned positive excluding prepayments, and a third if the multiple compresses further on no fundamental deterioration. The single-stock limit applies with force, because jurisdiction risk is not diversifiable within a China-focused book.
Selling criteria — four triggers
We would sell on any of: a price above our bull case of HK$684 without an earnings upgrade; advertising growth in single digits for two consecutive quarters; capex above RMB 50 billion a quarter into 2027 with net cash approaching zero and the buyback suspended; or a regulatory action that permanently changes the terms of monetisation.
The two catalysts that matter
Third-quarter results on 18 November 2026 will show whether capex held, whether free cash flow returned to positive and whether marketing services sustained growth above 15%. The FY2026 results in March 2027 will state next year's buyback intention alongside the capex plan. Restoring the buyback would signal the build is funded; a further reduction would confirm the balance sheet is doing the work.
23 — ConclusionA quality franchise priced for a bad outcome
The reason Tencent is priced as though it is in decline is not the business. It is that the company is spending heavily on AI without yet earning anything from it, while competitors with better models and comparable budgets do the same. Capex tripled, free cash flow went negative, the buyback was halved, and reported profit grew 0.7% because a non-cash line in the associate book moved. Read the accounts rather than the headlines and the core business compounds at a rate the multiple does not reflect.
What the framework adds is discipline. It does not ask whether AI is exciting; it asks whether the price compensates for the risk that AI fails. On the numbers it does: the operating business trades at 9.9 times forward non-IFRS earnings, return on equity is twice the cost of equity, cash conversion is 1.35 times, leverage is modest, and two-fifths of the price is an investment portfolio and a cash balance with nothing to do with AI. Our base case is HK$565, a 24.7% margin of safety; our bear case, HK$368, requires the franchise never to re-rate and the portfolio to halve. That is an asymmetric bet, and asymmetry rather than optimism is the reason to own the shares.
Constructive · HK$565 target
A quality franchise available at a price that assumes its largest investment programme earns nothing. We would become more constructive on disclosed AI revenue or a restored buyback, and we would turn negative if capex stays above RMB 50 billion a quarter into 2027 while advertising growth falls to single digits and net cash approaches zero.
24 — AppendixFinancial summary
| RMB millions unless stated | FY2023 | FY2024 | FY2025 | 2Q2025 | 2Q2026 |
|---|---|---|---|---|---|
| Revenue | 609,015 | 660,257 | 751,766 | 184,504 | 204,785 |
| Revenue growth, % | +10 | +8 | +14 | +15 | +11 |
| Gross profit | 293,081 | 349,246 | 422,593 | 105,013 | 118,433 |
| Gross margin, % | 48.1 | 52.9 | 56.2 | 57.0 | 57.8 |
| Operating profit, IFRS | — | 208,099 | 241,562 | 60,104 | 67,276 |
| Operating profit, non-IFRS | — | 237,811 | 280,656 | 69,248 | 75,636 |
| Profit attributable, IFRS | 115,216 | 194,073 | 224,842 | 55,628 | 56,022 |
| Profit attributable, non-IFRS | 157,688 | 222,703 | 259,626 | 63,052 | 68,415 |
| Basic EPS, RMB — IFRS | — | 20.938 | 24.749 | 6.115 | 6.207 |
| Basic EPS, RMB — non-IFRS | — | 24.027 | 28.577 | 6.937 | 7.581 |
| Return on closing equity, IFRS % | — | 19.9 | 19.5 | — | — |
| Operating cash flow | — | 258,521 | 303,052 | — | 52,700 |
| Capital expenditure | — | 76,760 | 79,198 | 19,107 | 52,784 |
| Free cash flow | — | — | — | — | −13,800 |
| Net cash | — | 102,422 | 107,145 | 74,592 | 58,191 |
| Weixin & WeChat MAU, m | 1,343 | 1,385 | 1,418 | 1,411 | 1,439 |
| Share repurchases, HKD bn | — | 112.0 | 80.0 | ~20 | 16.9 |
FY2023 gross profit and attributable profit are shown for continuity; the FY2023 equity base is not restated, so no FY2023 return on equity is presented. Free cash flow is disclosed quarterly and shown only where published on a comparable basis. Market data as of 13 October 2026.
25 — SourcesPrimary and secondary references
| Reference | Date | Used for |
|---|---|---|
| Tencent Holdings, announcement of 2026 second-quarter results (HKEXnews) | 12 Aug 2026 | 2Q2026 revenue, margin, profit, segment detail, capex, free cash flow, net cash, operating metrics, AI product disclosure, balance sheet |
| Tencent Holdings, announcement of 2025 annual results (HKEXnews) | 18 Mar 2026 | FY2025 revenue, segment revenue and margin, capex, operating cash flow, net cash, repurchases, final dividend, investee holdings, equity |
| Tencent Holdings, announcement of 2026 first-quarter results | 13 May 2026 | 1Q2026 comparatives; capex and net-cash run rate |
| Tencent Holdings, cash dividend announcement and AGM timetable | 18 Mar 2026 | HKD 5.30 final dividend, ex-date and payment date |
| HKEX next-day disclosure returns and repurchase filings | 2026 year to 9 Oct | Daily buyback executions and prices |
| Wind compilation of Hong Kong repurchase activity | to 9 Oct 2026 | HKD 32.0 billion of 2026 year-to-date repurchases |
| IMF World Economic Outlook | Jul 2026 | China real GDP growth forecasts and revisions |
| National statistics bureaux and market-data providers | 2026 | Retail-sales growth; share price, market capitalisation, 52-week range, peer multiples |
| Broker target-price compilations and consensus summaries | Aug–Oct 2026 | FY2026 revenue and earnings estimates; street reference points |
| Sell-side commentary on the Weixin AI agent | Sep–Oct 2026 | Xiaowei pilot status; the shift to an ecosystem-advantage narrative |
26 — DisclosureConflicts, positions and method
Positions and compensation. Farstar Capital holds no position in Tencent Holdings Limited (0700.HK) or in any company named in this report, and receives no compensation from Tencent or from any party with an interest in the securities discussed. Coverage is selected by us. This report was not previewed by the company.
Every figure traces to a filing, an earnings release, a company announcement, an international agency publication or a market data provider, each identified above; derived numbers are stated as such. The FY2026 estimates, the peer and historical multiples and the discounted cash flow are ours and are labelled indicative where they are not reported figures. Constructive means we expect a total return materially above the market over twelve months, with the target set at our base case. Tencent does not disclose revenue or unit economics for its AI products, nor the composition of its unlisted investment book, so our assessment of both is inferred — a limitation reflected in the width of our scenario range.
Not investment advice. This report is published for informational purposes only and is not an offer, a solicitation, or a recommendation to buy or sell any security. Figures are believed accurate as of the publication date but are not warranted, and past performance does not indicate future results. Readers should conduct their own diligence and consult a licensed adviser before acting.