Financebotresearch desk研究台

TME › analyze

TME · Analyze

ACCUMULATE Consumer

Price $8.66 · 52wk range $7.94-$26.70 · all-time range $2.95-$32.25 · MktCap (corrected) $14.18B

0. Knowledge check

python .mcp/kb.py find TME and Knowledge/INDEX.md returned no live note and no prior report on TME — this is a first-look analysis, nothing to cite or supersede. Given TME is a China-domiciled, dual-class-adjacent ADS with a non-USD reporting currency, three Playbook pitfalls were checked before trusting any vendor number, and two fired immediately on the raw fin.py/Yahoo payload:

  • [[pitfall-yahoo-share-count-dual-class-fpi]] — fired. See §1.
  • [[pitfall-currency-mixed-ev-is-invertible-on-adrs]] — fired. See §1.
  • [[pitfall-adr-ticker-resolution]] — checked, did not fire; TME resolved to the correct company (Tencent Music Entertainment Group) with a clean name match.

1. Data-quality gate — corrections applied before any analysis below

TME's true share count and enterprise value are both broken in the vendor payload, in ways this repo has documented before but not yet on this name. Both are fixed here using primary filings (SEC 6-K, Q2 2026 press release) and the inversion method from the cited pitfall, and the corrected figures are what the rest of this report uses.

Field Vendor (fin.py/Yahoo raw) True figure Source
sharesOutstanding 804.80M — Yahoo snapshot field
Balance-sheet "Ordinary Shares Number" 3.10B — pre-ADS-conversion total (2 ordinary shares = 1 ADS)
True ADS-equivalent share count — ~1.637B Class A 1,636.66M + Class B 1,640.46M (held by Tencent via Min River) ÷ 2, as of 6/30/26 6-K
Diluted wtd-avg (income stmt, used for EPS) 1.55B consistent with true count fin.py income statement
enterpriseValue $2.58B — Yahoo raw
enterpriseToEbitda 0.29x — Yahoo raw
enterpriseToRevenue 0.076x — Yahoo raw

Mechanism (both confirmed vendor failures, not new ones — see the cited notes for the general rule):

  1. Share count: sharesOutstanding (804.8M) captures roughly half of Class A alone and entirely omits Tencent's Class B stake (held via Min River, ~50.1% of the combined class, confirmed separately by a 55.5%-stake Schedule 13G/A). marketCap in the same payload is built off impliedSharesOutstanding (1,637,273,858) — which matches the primary-filing count almost exactly. MktCap ($14.18B) is therefore correct; per-share EPS/BVPS/Graham IV built off the diluted-EPS share count are correct; sharesOutstanding alone is the broken field.
  2. Enterprise value: financialCurrency: "CNY" is explicit in the raw payload — TME reports in RMB, trades in USD. EV is computed as USD MktCap − RMB cash/investment figure, unconverted. Inverting per the pitfall: local net cash = USD MktCap − reported EV = 14.18B − 2.58B = RMB 11.60B. Converting at ~7.15 RMB/USD gives true net cash ≈ $1.62B and true EV ≈ $12.5-13.0B — not $2.58B. Rebuilt EV/EBITDA ≈ 9-10x (vs. the vendor's 0.29x) and EV/Revenue ≈ 2.6-2.8x (vs. 0.076x). This is the difference between "priced for liquidation" and "priced like a moderately-growing, net-cash, 40%-gross-margin platform business" — a completely different read of the same company.

Also flagged, lower-severity: Yahoo's raw totalDebt field ($16.88B) does not reconcile to any quarterly balance-sheet line item (Total Debt ranges RMB 3.8-6.1B across the last five quarters) — a ~3x divergence. The granular, filing-sourced quarterly figure is used throughout this report per [[principle-primary-source-beats-vendor]]. Separately, roic.ai's per-share/cash- flow data labels TME's currency as HKD, which is very likely a metadata mislabel (numbers reconcile to RMB within a few percent, not to actual HKD) — flagged, not solved; not material to any conclusion below since both sources land on the same figures once you don't trust the label.

2. Fundamentals (annual, RMB — TME's reporting currency; only 4yrs of statement history is

available from either fin.py or roic.ai's free tier, short of the framework's 5-8yr target — stated as a limit, not papered over)

Metric FY2022 FY2023 FY2024 FY2025
Revenue 28.34B 27.75B 28.40B 32.90B
Gross margin 30.9% 35.3% 42.3% 44.3%
Operating margin 12.5% 17.9% 26.2% 29.6%
Net income 3.68B 4.92B 6.64B 11.06B
EPS (diluted, ADS-equiv) 2.28 3.10 4.24 7.12
OCF 7.48B 7.34B 10.28B 10.23B
FCF 6.43B 6.17B 9.24B 9.04B
Diluted shares (ADS-equiv) 1.62B 1.58B 1.57B 1.55B
  • FCF 3yr CAGR (2022→2025): 12.1%. Revenue CAGR 5.1% (3yr) — most of the FCF growth is margin expansion, not top-line growth (see §3 moat — this is the crux of the whole thesis).
  • Earnings-quality flag: FY2025's net income jump (+66% YoY) is not purely operating. Quarterly income statements show Q1 2025 alone carried a RMB 2.46B "other income/expense" swing (vs. a normal ~RMB 0-150M quarterly run-rate) — most likely a mark-to-market gain on a strategic equity holding. Operating income summed across FY2025's four quarters is RMB 9.74B, consistent with a genuine ~31% margin improvement YoY — but reported net-income growth overstates the operating story by a wide margin. Treat the 44.3% 3yr net-income CAGR as optimistic; the FCF and operating-income trends are the more trustworthy read.
  • Capital allocation: buybacks every year (RMB 390M-3.13B/yr, 2022-2025), first dividend paid 2024 (raised at each subsequent declaration: $0.137 → $0.18 → $0.24 trailing), debt paydown ongoing (Debt fell from RMB 6.05B, 2023/24, to RMB 3.81B, FY2025). One large countervailing event: the Ximalaya acquisition (closed 2026-05-18, $2.6B/RMB 18.6B, cash + up to 175.3M new Class A shares) added back roughly a decade of buyback-driven share reduction in one transaction — true ADS-equivalent share count is now flat-to-up YoY for the first time in years once the new shares are counted. This is real dilution, not a data artifact.
  • Debt/Assets: ~3.7-4.8% across all recent quarters — negligible leverage; net cash position (true net cash ≈ $1.6B) that comfortably funded the Ximalaya cash consideration without new debt of any real size (Q1 2026 total debt RMB 4.84B, up only modestly from RMB 3.81B).
  • Per-share metrics (revenue/share ~RMB 21.2, FCF/share ~RMB 5.8 ≈ $0.82 USD) are usable directly — the diluted-share basis they're built on matches the true ADS-equivalent count (§1).
  • Dividend-grower overlay (partial — apply with caution): yield 2.73%, payout ratio ~29% (well covered by FCF), but only 2 years of payer history — too short for the framework's 8-10yr dividend CAGR test. Flagged as an emerging, well-covered income stream, not yet a scored dividend-grower thesis.

3. Moat

Quantitative base is genuinely strong and improving. Gross margin has expanded in a straight line for four consecutive years — 30.9% → 35.3% → 42.3% → 44.3% — the kind of monotonic, multi-year margin trend the Playbook treats as real moat evidence rather than noise ([[pattern-margin-expansion-is-a-finite-growth-lever]] is the relevant caution, addressed below). ROIC (roic.ai) rose from 9.9% (2024) to 14.1% (2025); ROE from 10.6% to 15.2%. Both trends are real, though FY2025's figures are inflated somewhat by the one-off other-income item noted in §2.

Where the margin came from. TME lost the right to exclusive music licensing in China's 2021 antitrust settlement — a structural moat weakening at the time. The subsequent margin expansion has come from (a) no longer overpaying for exclusivity, (b) mix shift toward higher-margin social entertainment/live-streaming and karaoke (WeSing), and (c) scale leverage on a largely fixed content-cost base. This is a finite lever, not an infinite one — the framework's own pattern note on margin expansion applies directly: management cannot keep re-running "we found more efficiency" as the growth story once the mix shift is largely complete, and Ximalaya's H2 2026 guided margin drag (see §4) is the first sign the lever may be running out for now.

Adversarial stress-test — how would a well-funded rival attack? - ByteDance/Douyin is the sharpest threat, not a hypothetical new entrant. Short-video is already where a large share of Chinese users discover music before ever opening a dedicated streaming app — an attention-layer disintermediation risk that doesn't require Douyin to build a competing music subscription product, only to keep winning the discovery layer. This is the same mechanism the Playbook's shrinking-base pattern describes: casual, low-intent listening migrates to the free, bundled surface first. - NetEase Cloud Music is a real, named competitor — 2026 coverage explicitly cites "heightened competition in China's music streaming market" as a drag on TME's growth, not a latent risk. - Global streamers (Spotify, Apple Music) are structurally locked out of mainland China by policy — an unusual case where regulation protects the incumbent rather than attacking it. - The strongest, most durable pillar is the Tencent super-app embed — WeChat/QQ social-graph reuse, payment rails, and cross-promotion across Tencent's ecosystem are not something a standalone rival can replicate without an equivalent super-app, and Tencent's own dominance there is not in question. This is the one moat source that looks close to permanent.

Regulatory ceiling, not just a rival — and it just got tighter. China's regulator approved the Ximalaya deal in May 2026 with five conditions, including that the combined company not raise prices, not reduce the proportion of free content, and not enter exclusive licensing. That directly caps three of the standard monetization levers (price, paywall depth, exclusivity) for the combined entity going forward — a real, name-specific ceiling on how much of the moat can be monetized, independent of how strong user engagement is.

Evergreen assessment: music/audio consumption itself is close to a permanent human behavior — genuinely evergreen at the category level. Whether it stays dedicated-app-centric rather than increasingly bundled into short-video and super-app feeds is a live 5-10yr question, not an existential one on a 1-2yr view, but real enough to keep conviction capped.

4. Sentiment — why is it near a 52-week low right now

TME is not quietly drifting near its low; it got there through two distinct, dated sell-offs five months apart, and the reasons are different each time — that matters for whether this is cyclical fear or structural decay.

March 2026 — the first crash (~-25-33% over two sessions). Q4 FY2025 results showed solid revenue growth, but the stock cratered on subscriber-growth deceleration — the market re-rated TME from "still-compounding platform" toward "maturing business, hot-growth days behind it," per contemporaneous coverage. Music-subscriber growth had been decelerating for several quarters by this point.

May 2026 — the Ximalaya acquisition closes ($2.6B, cash + up to 175.3M new Class A shares, approved with the price/free-tier/exclusivity conditions in §3). Framed by management (Executive Chairman Cussion Pang) as a long-form-audio bet to compete more broadly with global audio platforms (Spotify's playbook), not a defensive move.

August 2026 — the second crash (-9-12% around the Q2 print, Aug 11-13). Revenue beat ($1.32B, +Q2), but: (a) subscriber growth continued decelerating to ~8%, "more robust in past quarters" per the reporting; (b) gross margin compressed slightly (44.2% vs. 44.4% YoY) on revenue-mix shift; (c) opex rose to 14.5% of revenue (from 13.7%) on regulatory/compliance costs; (d) management explicitly guided H2 2026 margins lower, citing Ximalaya's own costs and revenue mix — the company disclosed this was not fully apparent at the time of the acquisition, which reads as a real underwriting miss, not conservatism; (e) TME simultaneously completed a $400M buyback (-2.82% of shares) while flagging the margin deterioration — returning cash into softening near-term profitability, which the market read as tone-deaf rather than confidence-signaling.

Structural overhang, not name-specific: broader resistance to US-listed Chinese ADRs (audit-access and delisting-risk sentiment) sits on top of all of this and is not something TME's own execution can fix — worth pricing as a standing discount, not a resolvable catalyst.

Institutional / insider read: institutions hold 60.6% (Schroder, BlackRock, Invesco, Krane, Vanguard among the top ten) — a broad, credible base, not a thin retail-driven float. Short interest is moderate (5.6% of float, 3.76 days to cover) — not a crowded short. Yahoo's insider transaction feed returned empty and heldPercentInsiders: 0 — expected, since Tencent's control stake sits in Class B ordinary shares, outside Yahoo's "insider" classification for this structure, not a genuine golden- or red-flag insider signal either way.

Read: cyclical-with-a-real-execution-wrinkle, not structural collapse. The core China music-streaming franchise is intact and still gross-margin-expanding; the acquisition thesis (long-form audio) is unproven and management's own guidance says the next 1-2 quarters will be worse before any Ximalaya synergy shows up. That is a genuine, undigested risk — not merely sentiment overreacting to a good business.

5. Valuation

Applied conditionally: TME is a maturing-growth platform with a nascent, low-payout dividend — Graham and FCF-based approaches carry the weight; DYT/DDM are included for completeness but explicitly discounted given a 2-year dividend history and a 29% payout ratio (most of the return case is earnings growth and buybacks/M&A, not the dividend).

Model Inputs Output Weight
Graham IV EPS(ttm) $0.84, BVPS $7.13 → √(22.5×0.84×7.13); fwd EPS $1.00 → √(22.5×1.00×7.13) $11.6 (ttm) – $12.7 (fwd) High
FCF multiple FCF/share ≈ $0.82 (FY2025); 12-15x FCF band for a 40%+ gross-margin, net-cash, mid-single-digit-to-low-teens grower $9.8 – $12.2 High
True EV/EBITDA rebuilt EV ~$12.5-13.0B ÷ EBITDA ~$1.25-1.35B (RMB 8.9-9.7B ÷ ~7.15) ~9-10x — moderate, not a statistical-cheap outlier once corrected Medium (cross-check)
Bogle expected return div yield 2.7% + earnings growth 8-12% (conservative — normalized ex-one-off, revenue growth + buyback tailwind, net of guided margin drag) ± 0% multiple change (no re-rating assumed) ~11-15% Medium (broad context)
DYT 2.73% yield vs. no real historical band (payer since 2024 only) Not informative — too short a history Low
DDM Gordon growth, D $0.24, g 6%, r 10% → $6.36 Undervalues the name — low-payout dividend, most value accrues via earnings/buybacks, not the payout Low

Fair value range: $10-13, central estimate ~$11.5. Current price $8.66 sits roughly 15-30% below that range — a real but not extreme discount. This is meaningfully better than the vendor's broken EV multiples suggest ("practically free" at 0.29x EV/EBITDA is fiction), but it is also not the screaming statistical bargain a naive read of the 52-week-low framing implies. Trim as a forward multiple, not a fixed dollar: current forward P/E is 8.6x; set Trim at 16x forward — below the ~20-24x the stock commanded during its 2024-25 re-rating peak, reflecting the now-known margin/competitive/regulatory ceiling, but still leaving real room before the position should be trimmed on valuation grounds alone.

6. Synthesis — weighted verdict

Two things are true at once, and the tension is the whole thesis. Fundamentals + the corrected valuation say this is a genuinely profitable, net-cash, margin-expanding platform priced at a real (if moderate) discount to fair value — a legitimate "quality-on-sale" setup once the vendor's EV and share-count errors are stripped out. Sentiment says the market has now been right twice in five months about deteriorating near-term operating momentum (subscriber deceleration in March, margin-guidance cut in August), and management's own words say the next 1-2 quarters get worse before Ximalaya's contribution is even visible. Weighting: for a maturing-growth platform (neither a mature dividend cash-cow nor a pure high-growth story), Fundamentals and Sentiment are weighted roughly equally here, with Moat supplying the structural ceiling (real Tencent-ecosystem lock-in, but capped monetization under the antitrust conditions) and Valuation confirming there is a margin of safety without pretending it is a statistical outlier.

Named tensions: 1. Margin trend vs. guided reversal — four years of gross-margin expansion is real moat evidence, but management itself says H2 2026 breaks the trend on Ximalaya drag. Believing the long-run trend resumes requires believing this is a temporary integration cost, not a new lower-margin steady state. 2. Cheap-on-corrected-multiples vs. two guidance misses in five months — the valuation case is sound only if you trust the next print; this management has twice recently under-called its own near-term trajectory (subscriber growth, and explicitly admitted the Ximalaya margin impact "was not disclosed at the time of acquisition"). 3. Durable ecosystem moat vs. capped monetization — the Tencent super-app embed looks close to permanent; the regulator has simultaneously removed price, free-tier, and exclusivity as levers for extracting more value from it. A durable moat with a capped ceiling is a real thing, and it argues against paying up even if the franchise itself is safe.

Verdict: ACCUMULATE, conviction 6.0 — a close call, not a high-conviction call. The data-quality corrections turn this from "screening as free" (vendor fiction) into "modestly undervalued, real business, unresolved near-term risk" (the honest read). That combination argues for scaling in gradually into the current $7.50-9.50 zone rather than either avoiding it (the balance sheet and margin history are too good to file as a value trap) or sizing up heavily before the Q3 print resolves whether the Ximalaya drag is temporary. Recheck at the Q3 FY26 print (~Nov 11 2026) — the first full quarter that should show the guided margin impact, which is the single number that moves this from ACCUMULATE toward either BUY (drag proves transitory) or WATCH/HOLD (drag proves structural).

Risks (the 1-2 things that break the thesis)

  1. Ximalaya's margin drag is structural, not transitory. If long-form audio content costs permanently reset the consolidated gross margin lower — rather than integrating toward TME's own margin profile within a few quarters — the four-year margin-expansion trend (the core moat evidence in this report) reverses, and the valuation case built on FY2025 margins was built on a peak, not a new floor.
  2. China-ADR structural risk (audit access, delisting sentiment, RMB/USD policy shifts) is a standing discount this report cannot underwrite away. It sits on top of the operating thesis regardless of how well TME itself executes, and it is the kind of risk that resolves on a political timeline, not an earnings one.