Rakuten Group4755.T
The consolidated accounts say almost nothing true about Rakuten. Net "cash" of ¥11.9tn is a banking balance-sheet artefact ; the P/E reads on a negative number. Valued part by part, the fintech alone — a listed online bank, an 85%-owned card book, a broker — is worth most of the market cap, and the sum of the parts lands above the price. What decides the case is narrower and more uncomfortable : whether the ordinary holder, standing behind ¥480bn of perpetual hybrids, the holding debt and a 2027 refinancing wall, captures that value. At ¥764 the weighted asymmetry has already compressed to +5.7%, against a −50.6% permanent-loss tail.
The three regulated financial subsidiaries, valued at Rakuten's economic stake — Rakuten Bank at 49.94% of its ¥941bn listing, Rakuten Card at 85% of a ¥825bn implied value, Securities and the smaller fintech below that — are worth roughly ¥1,452bn attributable.
Internet Services, a mature marketplace at 10× its ¥89bn segment operating profit, adds ¥889bn ; Mobile, on enterprise-to-sales net of its dedicated debt, adds ¥240bn. Gross, the six parts come to ¥2,581bn.
Strip ¥300bn of non-financial holding debt and half the ¥480bn hybrid stack as quasi-debt, apply a 7.5% subordination discount, and attributable equity is ¥1,888bn in the base case.
The market capitalises Rakuten at ¥1,661bn. The parts sum above the price — what reaches the ordinary holder is what survives the hybrids, the holding debt and the 2027 wall.
The useful thing to understand about Rakuten is that its headline financials are not just uninformative, they are actively misleading. The group reports ¥11.9tn of net "cash", which sounds like a fortress and is in fact the deposits sitting inside Rakuten Bank. Consolidated free cash flow swings by trillions from year to year on the movement of those deposits and its securities book. Net income is negative, so the P/E is meaningless — the "54.5x forward" the screens show is a FY2027 number that only exists because FY2026 earnings are still forecast below zero. None of the aggregates an investor normally reaches for can be trusted here. The only way to see the company is to take it apart.
Taken apart, it is a financial holding company wearing an e-commerce name. One segment, fintech — the card, the bank, the broker — earned ¥199.9bn of operating profit in FY2025 at a 20.5% margin, which is more than twice what the marketplace earns. A second, Mobile, lost ¥161.8bn, most of it now the depreciation of a mobile network that is already built and paid for. A third, Internet Services, is a mature, mildly profitable marketplace. Stack those three and you get a consolidated operating line of ¥14.4bn that describes none of them. The value lives in the fintech ; the risk lives in Mobile and in the balance sheet ; and the consolidated number hides both.
So the operational question — has the turnaround happened? — is largely settled. Managerial operating profit rose roughly fifteen-fold to ¥106.3bn, Mobile crossed into positive EBITDA, the group's IFRS operating line turned positive in Q1 FY2026, and the capex that built the network has collapsed to 2.6% of revenue from a 15%-plus peak. The market has watched all of this and re-rated the price-to-book to 3.15×, a decade high, up more than 200% since the Mobile launch. The recovery is not an ignored bargain. It is a thesis the price has already largely paid for — and the −27% year-to-date is that multiple beginning to give some of it back.
The debate that is not settled is who captures the value. Rakuten Bank is separately listed and only 49.94% owned ; minorities took ¥54.7bn of profit in FY2025. Rakuten Card sits at 85%. Ahead of the ordinary shareholder stand ¥479.7bn of perpetual hybrids that rank senior, a reconstructed ¥300bn of non-financial holding debt buried inside the banking balance sheet, and a 2027 maturity wall carrying dollar senior paper at an 11.25% coupon. The parts are real and they sum above the market cap. The question is how much of that reaches an owner who is, in capital-structure terms, a junior creditor on a crystallisation that has not happened yet.
The position framing is observation with a documented long bias, not ownership at this level. The sum-of-the-parts base case sits +13.7% above spot, but the probability-weighted fair value is only +5.7%, against a bear case that loses more than half — a 0.27× asymmetry. Conviction is moderate. The dossier becomes a long on two conditions the price does not yet hold : a materially lower entry that restores the asymmetry, and evidence that the October 2026 fintech reorganisation and the 2027 refinancing resolve in the owner's favour. Both are on an observable calendar.
The decade reads as three regimes with a single decision at their hinge. Through 2018 Rakuten was a profitable e-commerce-and-fintech compounder, operating margin in the mid-teens, return on equity respectable, dividend steady. Then in April 2019 it chose to build a fourth national mobile network from scratch — the most capital-intensive route available — and spent the next four years in a telecom capex war that took the operating line from +¥175bn to −¥360bn at the 2022 trough and diluted the share count by more than half. The third regime, since 2024, is the accounting of the exit : capex has fallen off a cliff, the Mobile loss is narrowing, managerial profit has recovered sharply. The reason a ten-year average multiple is useless here is that the middle of the series is a crater, not a trend.
| Inflection | FY 2015Pre-Mobile | FY 2018Peak | FY 2022Mobile trough | FY 2024Inflection | FY 2025Recovery |
|---|---|---|---|---|---|
| Revenue (¥bn) | 713.6 | 1,101.5 | 1,927.9 | 2,279.2 | 2,496.6 |
| Operating profit — IFRS (¥bn) | 94.7 | 175.2 | −359.7 | 55.3 | 14.4 |
| Operating margin | 13.3% | 15.9% | −18.7% | 2.4% | 0.6% |
| Net income attributable (¥bn) | 44.4 | 142.3 | −372.9 | −162.4 | −177.9 |
| Capex (¥bn) | 19.7 | 23.4 | 298.7 | 84.0 | 65.7 |
| Depreciation & amortisation (¥bn) | 40.1 | 72.4 | 252.2 | 316.4 | 320.5 |
| Return on equity | 6.7% | 19.5% | −60.3% | −31.0% | −18.5% |
| Shares outstanding (m) | 1,375 | 1,350 | 2,142 | 2,154 | 2,170 |
| Perpetual hybrid capital (¥bn) | 0 | 0 | 317.3 | 398.7 | 479.7 |
Source: data pack 1 July 2026 and issuer disclosure ; FY-December basis, no stock split. IFRS operating profit is genuinely misleading in the recovery years : FY2024 carries a ¥106.9bn AST SpaceMobile remeasurement gain absent in FY2025, so the FY2025 IFRS line (¥14.4bn) reads below FY2024 even though managerial Non-GAAP operating profit rose to ¥106.3bn. Net income attributable stays negative on network depreciation and recurring impairments. Consolidated FCF and net debt are omitted here as banking-balance-sheet artefacts.
Three consequences carry into the thesis. The equity holder financed the entire Mobile experiment and has been paid nothing for it, so any valuation has to run through the hybrid stack and the dilution, not around them. The revenue growth was real but bought at the price of profitability, which means head-line growth is a false signal and only segment-level economics matter. And the one durable, mechanical fact in the whole record is the capex cliff — capex down 78% from its peak against depreciation running roughly five times higher — which is why the industrial cash flow can turn before the accounting profit does. That last point is the load-bearing beam of the bull case, and the one item that does not depend on a narrative.
The engine only makes sense at the segment level, because the three businesses have almost nothing economically in common. FinTech earned ¥199.9bn of segment operating profit in FY2025, on a 20.5% margin — a regulated financial spread book, Rakuten Bank running a 21.7% standalone return on equity. Internet Services earned ¥88.9bn at roughly 6.5%, an ordinary mature marketplace. Mobile lost ¥161.8bn. The consolidated 0.6% IFRS margin is the weighted average of a genuine bank, a middling retailer and a cash pit, which is exactly why a single group multiple is the wrong instrument.
What matters about the Mobile loss is its composition, because it is not what it looks like. Of the ¥320.5bn of group depreciation, ¥188bn — about three-fifths — is the amortisation of the Mobile network, a sunk asset that no longer consumes cash. Mobile's adjusted EBITDA turned positive in FY2025 at +¥28.8bn, its first, and the operating loss is on a declining path (−¥161.8bn shrinking toward the low tens of billions in the base plan). The live risk is not the depreciation, it is the subscriber economics : average revenue per user is around ¥2,442 a month against a break-even band of ¥2,500–3,000, so each marginal subscriber above ten million is still, for now, added slightly below the line rather than above it.
The capex story deserves one correction the desk audit forced. The headline "depreciation is five times capex" overstates the cliff : that ratio compares full depreciation, including amortisation of lease-financed assets, against the narrow cash-capex line. On total capital investment of ¥287bn, depreciation is 1.12× — meaningful, but not the near-total run-off the raw ratio implies. The FY2026 plan itself is running above ¥200bn, roughly ¥87bn of it deferred FY2025 spend, so part of the FY2025 collapse to ¥65.7bn was a calendar underspend rather than a purely structural drop. The industrial free cash flow still turns — normalised at roughly ¥215bn — but it turns less mechanically than the raw cliff suggests.
Everything then routes through a balance sheet built against the ordinary holder. The ¥479.7bn of perpetual hybrids rank senior and accrue distributions the equity does not ; the real non-financial holding debt is buried inside the banking balance sheet and has to be reconstructed by stripping out Bank, Card and Securities ; and a 2027 maturity wall carries dollar senior paper at an 11.25% coupon, a direct read on how expensive holding-company funding has become. The subordination is the whole game : the parts can be worth more than the market cap and the equity holder can still be left with a junior claim on a crystallisation that has to survive the 2027 wall first.
This pillar carries the thesis because it decides whether the equity holder ever earns a positive return on capital, and today it is structurally negative — consolidated return on equity was −18.5% in FY2025, the sixth loss-making year in a row. Underneath sits a real sub-model : the fintech book compounds a regulated spread and Rakuten Bank runs a 21.7% standalone return on equity. But the group model does not yet convert that into an attributable return, because Mobile still absorbs it, the capex plan is climbing back above ¥200bn, and the holding structure taxes the owner. Genuine assets, a broken group return.
The second cardinal, because the entire question is whether value reaches the ordinary holder, and the capital structure is built the other way. The decade added nearly 60% to the share count, stacked ¥479.7bn of perpetual hybrids senior to the equity, and suspended the dividend in FY2023 — shareholder yield is zero, with no buyback and no floor of return. The one improving thread is the October 2026 fintech reorganisation, which can lift Rakuten's economic interest in the Bank from 49.94% toward 72.35% and so raise the attributable value. That reorganisation is the single governance item working for the owner rather than against.
Balance-sheet demand rather than transactional : fintech balances grew +19.0%, Mobile passed ten million subscribers, the membership base is captive. Commerce is mature (+6.8%) and the whole book is levered to the Japanese credit cycle and Bank of Japan rates.
The only Japanese super-app to own a bank, a card, a broker and a mobile network at once, cross-sold off a national membership base at near-zero acquisition cost. But the individual verticals are contested and Mobile's ARPU still sits below break-even ; the ecosystem is broad, not deep.
Strategic nerve and a credible recent execution of the inflection — managerial operating profit up fifteen-fold, Mobile to positive EBITDA. Set against a decade record of value destruction : the in-house MNO choice, poorly controlled international dispersion, recurring impairments dressed as one-offs.
A turnaround with real assets and a broken model, not a compounder — well below the two quality names in the same bucket (Zozo and Mercari at ~17/25). The two cardinals both score 2.0, which is the grade telling you where the case lives : not in the quality of the businesses, which is mixed, but in whether a deep-value sum-of-the-parts crystallises through a hostile capital structure. The score is consistent with the read — assets worth owning, held by an owner who has been paid nothing to wait.
Does the equity holder capture the fintech value, or only a diluted fraction of it ?
Is the inflection a re-rating to come, or already in a decade-ceiling price ?
The operational inflection is genuine and freshly booked — managerial operating profit up fifteen-fold, Mobile EBITDA positive, IFRS operating profit positive in Q1 FY2026. But the price-to-book is 3.15×, a decade high against a 2018 trough of 1.28×, up more than 200% since the Mobile launch. The recovery the market can see is largely paid for ; the −27% year-to-date is that multiple reverting. This is why the bucket's first rule — 08a prices the second derivative, not the level of quality — leaves no easy re-rating on the table without a fresh catalyst.
Is the capex cliff structural, or a calendar underspend that reverses ?
Capex fell to 2.6% of revenue from a 15%-plus peak, which reads as a permanent turn in the industrial cash flow. Part of it is not. The FY2026 plan runs above ¥200bn, roughly ¥87bn of it FY2025 spend deferred, which is the explicit trigger of the inherited red flag : Mobile capex climbing back over ¥100bn weakens the "cliff" thesis. The industrial free cash flow still normalises around ¥215bn — a 12.9% yield on the current capitalisation — but on a proxy for lease neutralisation that cannot be validated outside the annual filing.
At ¥764 and a decade-high 3.15× price-to-book, the market is pricing a substantial turnaround success — Mobile ceasing to destroy capital and a fintech worth more than the consolidated line shows. The behavioural tell is the price-to-book re-rating of more than 200% since 2019 : the "free Mobile" sum-of-the-parts is a thesis the market has largely adopted, not an ignored bargain, and the −27% year-to-date is that anticipation partly reversing. Every consolidated multiple is rejected here — the P/E on a negative number, EV/EBITDA on a banking net-cash artefact, FCF yield on deposit flows — so only the attributable sum-of-the-parts carries a signal. What is not in the price is the crystallisation of the October reorganisation or a durably profitable Mobile. On the wired parts, base-case attributable equity of ¥1,888bn sits +13.7% above spot, but the probability weighting pulls the fair value back to +5.7%.
The 2027 refinancing wall reprices at the 11.25% senior coupon and cannot be placed at a sustainable cost ; the Mobile-related debt is not offset and lifts net holding debt toward ¥600bn ; the reorganisation stalls at 49.94% and the Bank de-rates. This is a permanent loss, not a timing wobble — the wall destroys equity value irreversibly, with no restoration path on a modellable horizon until it is cleared. The floor at ¥378 is the fintech assets at their un-reorganised stake, net of the hybrids.
The October reorganisation executes cleanly, Mobile holds its positive EBITDA, the hybrids are treated at half credit and the holding debt at ¥300bn. The parts — Bank at 49.94% of its listing, Card at 85% of an ¥825bn value, Internet at 10× segment profit, Mobile on enterprise-to-sales — sum to ¥2,581bn gross, ¥1,888bn of attributable equity after the bridge and a 7.5% subordination discount. A consolidated re-rating is not required ; the fair value does not need one.
The un-priced levers fire together. The reorganisation crystallises the Bank at 72.35% economic and the listing re-rates ; Mobile turns operationally profitable and takes a growth multiple ; Internet re-rates to 13× ; the holding debt is worked down and Card is marked at its full Mizuho-implied ¥1.1tn. The path needs both the crystallisation and a profitable Mobile, neither of which is signalled today.
| KPI | Latest value | Status | What it tells us |
|---|---|---|---|
| FinTech segment operating profit | ¥199.9bn FY2025 | Cardinal | The value engine and the number-one lever (±12% of normalised earnings). A drift below ~¥170bn cuts fair value ~¥95/share and signals the credit cycle turning against the 20.5% margin. |
| 2027 refinancing spread | 11.25% senior coupon | Trigger | The bear mechanism. A placement spread above 11.25% at the 2027 wall flips the case to the bear ¥378 — the primary thesis breaker and analytical stop. |
| Mobile ARPU vs break-even | ¥2,442 / mo | Watch | Below the ¥2,500–3,000 break-even band. Sustained below ¥2,400 across two quarters pushes the Mobile turn out and flips toward the bear. |
| Mobile adjusted EBITDA | +¥28.8bn FY2025 | Inflection | First positive year. Durability across a full cycle, not one seasonal quarter, is what would let Mobile carry more than a distressed enterprise-to-sales multiple. |
| FinTech reorganisation (Oct 2026) | Bank 49.94% → 72.35% | Trigger | The attribution lever. Completion lifts attributable value ~¥200bn and opens the bull path ; a stall removes the bull uplift and leaves the discount uncrystallised. |
| Non-financial holding debt (proxy) | ~¥300bn Base | Watch | The most discriminant bridge line — ±¥207/share between a ¥300bn base and a ¥600bn bear. Rising blended cost of debt (2.5% FY2025 vs 0.5% FY2024) is the early signal. |
| Card cost of risk | Untested in downturn | Watch | The 20.5% fintech margin has never crossed a Japanese credit recession. +300bps of provisioning cuts fair value ~¥55/share ; reversible, but unproven. |
| Price-to-book vs decade corridor | 3.15× | Reference | At the decade ceiling (trough 1.28× in 2018, mean ~2.5×). The re-rating is largely done ; compression toward the mean on intact assets is what would restore the entry asymmetry. |
The case turns to a long if the crystallisation arrives or the entry improves. Completion of the October 2026 reorganisation at the 72.35% economic interest in Rakuten Bank, confirmed against the listed capitalisation, alongside a Mobile break-even that holds across a full year rather than one quarter, would move the dossier from watchlist to long and open the bull path toward ¥1,348. A price that falls materially — compressing the decade-ceiling multiple back toward its mean on unchanged assets — would do the same by restoring the asymmetry the current level does not offer. Either is observable ; neither is signalled today.
The case turns negative on the refinancing. A placement spread above the 11.25% senior coupon at the 2027 wall, or a Mobile ARPU sustained below ¥2,400 across two quarters, resets fair value toward the bear ¥378 — and the bear here is a permanent loss, not a timing disappointment, because the wall destroys equity value with no restoration path until it is cleared. A stalled reorganisation held at 49.94% would not break the thesis but would strip out the bull and flatten the asymmetry further.
The tail to watch most carefully is the one that ends the thesis outright. A Rakuten Bank capital ratio below 8% after the Class A conversion, combined with Mobile-related debt that cannot be refinanced at the 2027 wall, would crystallise a permanent loss and liquidate the sum-of-the-parts architecture the whole case rests on. It is the low-probability outcome the ¥479.7bn hybrid stack and the holding debt are built to survive — and the reason the equity, not the enterprise, is where the risk sits. Currently not signalled.
The information provided on this website is for informational and educational purposes only and should not be construed as financial, investment, legal, or tax advice. All content reflects the personal opinions, interpretations, and analyses of the author at the time of writing and is subject to change without notice. Nothing contained herein constitutes, or should be interpreted as, a recommendation, solicitation, or offer to buy or sell any securities, financial instruments, or other investment products. The author is not a licensed financial advisor, broker, or investment professional. Any references to specific assets, markets, or strategies are illustrative in nature and do not constitute personalized investment advice. Investing in financial markets involves risk, including the potential loss of capital. Past performance is not indicative of future results. Readers are solely responsible for their own investment decisions and should conduct their own independent research and due diligence before making any financial commitments. You are strongly encouraged to consult with a qualified financial advisor, legal professional, or other relevant specialist before making any investment or financial decisions. By accessing and using this blog, you agree that the author shall not be held liable for any direct or indirect losses, damages, or consequences arising from the use of, or reliance on, the information presented herein. All content is provided "as is" without any warranties of completeness, accuracy, or reliability.