Fuji Media Holdings (4676): a vault you can price but cannot open

I like to start where a shopkeeper would: what does this outfit sell, who pays, and why do they keep paying? Fuji Media Holdings is really two businesses wearing one hat E1. One is a television network — Fuji Television — that sells advertising slots to companies who want to reach a Japanese living room E4. The other is a Tokyo-and-Osaka landlord with hotels attached, collecting rent on office buildings, condominiums, and resorts E1E15. The law calls the whole thing a "certified broadcasting holding company" E2 — a fancy way of saying the government lets you hold a broadcast license and park a real-estate company underneath it.

I can understand both halves — I've owned newspapers and I've owned rental property — so I won't hide behind "too hard." But understanding a business and wanting to own it are two different things, and this is a fine place to teach a student the difference.

The television half is a melting ice cube. Last year it lost money at the segment line — an operating loss of ¥30,835M F78 — and the network on its own lost ¥32,515M F84. Broadcast and media revenue fell better than a quarter, down 27.4% F103. Part of that is the long tide going out on broadcast television everywhere; people watch their phones E4. But a good part is self-inflicted: a human-rights and compliance scandal at Fuji Television, after which advertisers walked E3.

Here is the lesson worth the price of admission. Years ago I bought American Express during the salad-oil scandal, and the whole point was that cardholders kept using their cards right through the mess — the franchise never flinched, so the scandal was a chance, not a wound. I run every scandal-hit business through that same test, B38. Fuji fails it. When the trouble came, the advertisers left, revenue collapsed, and the segment went red F103F78. That tells you what you need to know: this is not a franchise with a moat that holds under fire. It is an advertising-dependent business E4 whose customers could and did abandon it. The AmEx test isn't a formality — it's the whole difference between a wound and a wound that heals.

The real-estate half is the opposite: a genuinely good business. It earned ¥25,185M F80, it's growing on the back of condo sales and inbound tourism E15, and the company carries its rental property at ¥410,530M of book against a fair value of ¥476,114M — about ¥65,584M of value hiding in plain sight F89F88D3. There's also land at ¥388,581M of cost F31 that in Tokyo is likely worth more than the ledger shows — though I'll count only what's actually appraised in front of me, not what I'd like to imagine.

So we have a good landlord carrying a money-losing broadcaster. That can still be a fine investment — at the right price. And here is where I put my pencil down.

Two things happened to this balance sheet that a Graham man cannot ignore. First, the reported "profit" is a mirage. The company earned ¥6,499M to owners F15, but only because it sold cross-shareholdings for a ¥50,021M gain F40E13. Strip that out and the operating business lost ¥8,766M F36, and cash from operations was actually negative — minus ¥341M F29. When a company's only earnings come from selling the furniture, I don't call that earnings; that's B46, and it is Sanborn Map in reverse — investment gains papering over an operating decline.

Second — and this is the one that decides it — an activist pushed management into repurchasing ¥249,045M of stock F60, and they did it with borrowed money E11E12. Interest-bearing debt roughly tripled, to ¥615,134M F58; net assets fell from ¥830,023M to ¥561,467M F69F49; and the equity ratio dropped from 56.8% to 37.3% F22F23. Now, I've nothing against buybacks — but only when the cash is truly surplus and the price sits below conservative value. That is B73, and this deal failed both halves: they borrowed to do it, at a price around book. In the process they spent the one thing that would have made this interesting — the discount.

Look at the price. At ¥3,996 the stock is 1.04 times book D2 — a hair over, not under. Add back the after-tax real-estate gain and my honest appraisal of net asset value comes to roughly ¥593,000M against a market value of ¥569,213M D1F50D3 — call it a 4% discount. That is not a margin of safety; that is B93, and it fails cold. When I bought Commonwealth Trust I paid $50 for $125 of value. Three words, margin of safety — and this hasn't got one. Worse, half the "book" I'd be buying is a broadcaster losing ¥30bn a year F78, and its ¥457,175M of segment assets F81 aren't worth book if it keeps bleeding.

Would a lower price fix it? That's exactly why this stays on the watch list rather than in the reject pile. There is real Tokyo real estate here, and an activist forcing capital discipline and cross-holding sales — ¥48,595M sold this year against a target above ¥100bn F73E9E10. That is a genuine tailwind, and B41 nods to it. If this company ever separated its landlord from its television station, or if the price fell to two-thirds of a conservative appraisal — somewhere south of ¥2,800 — a student of Graham should look again.

But not today, and not here. At this price you'd pay full freight for a levered conglomerate whose profits are asset sales, whose broadcaster failed the only test that matters, and whose owners just borrowed to shrink their own equity. My verdict is watch — and the thing to watch is whether the broadcaster melts the net asset value faster than the price ever falls.

Invert first. Before I ask how Fuji Media wins, I write its obituary, and it writes itself. This is really a Tokyo real-estate developer E1 with a broadcaster chained to its ankle. The Media & Content segment lost ¥30,835M F78; Fuji Television alone lost ¥32,515M on its own books F84 as broadcast and media revenue fell 27.4% to ¥117,077M F103. The Urban Development & Tourism segment earned ¥25,185M F80 — and could not cover the broadcaster's bleeding. Consolidated, the operating result was a loss of ¥8,766M F36. The only reason the bottom line shows ¥6,499M of profit F15 is ¥50,021M of one-time gains from dumping cross-shareholdings F40E13. Strip the asset sales and the company lost money running its businesses. Substitute the plain word for the flattering one: this is not "earnings," it is a rummage sale.

Now the balance sheet, where the real mischief lives. Under activist pressure E8, management bought back ¥249,045M of its own stock F60E11 and paid for it by roughly tripling interest-bearing debt to ¥615,134M F58, ¥277,577M of it short-term F59E12. Net assets fell from ¥830,023M to ¥561,467M F69F49; the equity ratio dropped from 56.8% to 37.3% F22F23. They levered a loss-making operating company to hand cash to shareholders. Singleton bought back stock when it was cheap with money he did not need; this is the reverse — borrowing to shrink the equity of a business whose operating cash flow just went negative, −¥341M F29. Show me the incentive and I will show you the outcome: the buyback flatters per-share optics while the underlying engine deteriorates, and it was done to quiet agitators, not because the arithmetic demanded it. To their credit, the people running it are not looting the place — director pay is a modest ¥129M with no bonus taken in the loss year F102E18. That is the one clean thing here, and I note it, because I read the compensation before I read the strategy.

Name the moat's mechanism, not its existence. Two candidates. First, a broadcast license — Fuji Media is a certified broadcasting holding company operating by regulatory grace E2. But a regulatory moat around a segment that loses ¥30,835M F78 protects nothing worth protecting, and what the state grants after a human-rights scandal E3 the state can revoke or reprice. Second, real estate: irreplaceable Tokyo and Osaka land carried at ¥388,581M F31, rental property with a fair value of ¥476,114M against ¥410,530M of book F89F88E14 — a ¥65,584M unrealized cushion D3. That cushion is genuine. But leasing buildings is a low-return asset business — ¥23,162M of rent F90 against ¥410,530M of property, and the entire Urban Development segment earns ¥25,185M F80 on ¥673,812M of assets F82, under four percent. That is not a compounding machine; it is a savings account with tenants.

So what is being offered? A great business at a fair price beats a fair business at a great price, and this is neither half of that sentence. It is a mediocre operating business — ROE of 1.0% F25, an operating loss F36 — sitting on valuable but trapped assets. Trapped is the crux. The value-destroying broadcaster cannot be cut out. Management explicitly frames Fuji Television as "social infrastructure" that must keep transmitting through any revenue shock E6, and the whole holding-company structure exists to own it E2. The cancer-surgery formula — excise the folly, keep the wonderful core — is unavailable when the surgeon has vowed never to operate. You buy the raisins and the turds together, and you are told the turds are a public duty.

Then the price. The watchlist tag that brought me here — "below book" — is stale. At ¥3,996 the stock trades at 1.04× reported book D2, and reported book already carries securities at fair value F65. Re-derive value from the figures with no reference to any prior number: self-capital ¥546,719M F50, add the after-tax slice of the ¥65,584M property uplift D3, and a conservative net asset value lands near ¥590bn against a market capitalization of ¥569,213M D1. You are paying roughly full freight for a levered, loss-making conglomerate. The gap between price and conservative value is essentially gone; there is no margin of safety in that.

Count the forces pointing the same way, because the big effects come from combinations. Secular decline in terrestrial-television advertising E4; a reputation lost in a heartbeat and not yet earned back E3; an operating engine burning cash F29F36; a balance sheet freshly tripled in debt F58; and a loss-maker that structurally cannot be sold E6. That is a negative lollapalooza — five separately-survivable troubles that in confluence make a spiral, not a bargain. Against it sits one real positive, the land, and even that is now mortgaged to the buyback.

The verdict follows from the opportunity-cost hurdle, not from any yardstick of cheapness. Measured against simply doing nothing, this fails. It is a mediocre business whose one attraction — asset value — is fully priced and encumbered, whose moat guards its worst segment, and whose profit this year was a one-time sale. I do not need a precise sum-of-the-parts to know I do not want it. Pass. There are worse fates than sitting on your hands, and this is one of the situations that proves it.

Fuji Media Holdings sits across two businesses that could not be more different. The Urban Development and Tourism arm — office buildings, condominiums, and hotels spread across Tokyo and Osaka, plus a new Kobe aquarium resort — earned segment income of ¥25,185M on ¥193,495M of external revenue in FY2026 F80F79. The Media and Content arm — Fuji Television, Pony Canyon, streaming and film — lost ¥30,835M at the segment level F78. The broadcaster's standalone operating loss was ¥32,515M F84, and its broadcast advertising revenue fell 27.4% F103 after a human-rights and compliance affair at the subsidiary materialized as a real risk E3. Consolidated operating loss was ¥8,766M F36. Net income of ¥6,499M F15 exists only because the company sold ¥50,021M of cross-shareholdings in a one-time extraordinary gain F40E13. Strip that out, and the operating engine burned cash in FY2026.

So why did this land on a screen? Because at ¥3,996/share and a market cap of ¥569,213M D1, the stock looked like it might trade below book — price-to-book of 1.04× D2. And beneath the balance sheet there is real estate. Land carried at historical cost: ¥388,581M F31. Rental office buildings, condominiums, and commercial facilities in Tokyo and Osaka — the investment property footnote shows a book value of ¥410,530M and a fair value of ¥476,114M, a pre-tax unrealized gain of ¥65,584M sitting in the notes F88F89D3. Strategic cross-shareholdings still standing at ¥182,235M F71F72. Total assets ¥1,464,728M F47. A lot of stuff.

Now I do the stress test. This is where I always start — not with the upside story, but with what the floor actually looks like when I'm deliberately pessimistic. Cash and deposits at 90%: ¥75,839M from ¥84,265M F51. Short-term securities at 90%: ¥65,296M from ¥72,551M F52. Receivables and other current assets at 65%: ¥152,003M from ¥233,851M. Investment property at 70% of fair value: ¥333,280M from ¥476,114M F89. Other PP&E — broadcasting studios, equipment, owner-occupied facilities — at 40%: ¥82,152M from ¥205,380M. Investment securities at 65%: ¥241,996M from ¥372,302M F56. Remaining fixed assets — goodwill, equity-method, deferred tax — at 10%: ¥8,585M from ¥85,848M. Stressed total: roughly ¥959,000M.

Now subtract total liabilities: ¥903,260M F48.

Stressed floor: approximately ¥56,000M. Against a market cap of ¥569,213M D1, that is roughly 10% of the purchase price surviving a real stress. Even in a generous version — investment property at 80% of fair value, securities at 80% — the floor reaches perhaps ¥160,000M, barely 28% of market cap. This is not "heads I win, tails I don't lose much." This is "heads I win, tails I lose 70–90%." An equity cushion of ¥561,467M F49 is not a downside floor: it is book net assets sitting on top of levered real estate and a franchise losing money. P1 fails.

How did the balance sheet reach this condition? Under activist pressure, the company executed a ¥249,045M treasury buyback in FY2026 F60 — funded almost entirely by borrowing. Short-term borrowings rose to ¥277,577M F59, long-term debt to ¥315,557M F74, bonds ¥22,000M F75, for an interest-bearing total of ¥615,134M F58 E12. Net assets collapsed from ¥830,023M F69 to ¥561,467M F49. The equity ratio fell from 56.8% to 37.3% F22F23. Net debt now stands at ¥530,869M D4.

Here is the leverage problem, stated plainly. Operating cash flow was negative ¥341M F29. Interest paid was ¥3,898M F42. Cash on hand is ¥84,265M F51. And ¥277,577M of short-term borrowings F59 must be refinanced within the year. The company cannot pay down that principal from operations — it depends on rolling the bank facilities, on continuing to sell cross-shareholdings (targeting cumulative ¥100bn by FY2027 E9, having sold ¥48,595M this year F73), and on rental income of ¥23,162M F90 from the real estate book. Management explicitly frames Fuji Television as "social infrastructure" requiring a strong balance sheet and ample liquidity to keep broadcasting through prolonged revenue shocks E6 — which is an honest acknowledgment that the broadcaster cannot fund itself. The Dhandho survival question is: can this company survive two to three bad years without needing capital markets? The answer is no. P20 fails.

I have studied the FY2027 guidance: revenue ¥625,700M, operating income ¥40,100M, net income ¥26,100M F98F99F101 E21. I have read the Reform Action Plan, the Group Vision 2026-2030 Ver.1.0, the new Nomination and Compensation Committee established in June 2025 E16E17, the Human Rights Committee E17 E7. Management is doing the right things. But I am not buying management effort — I am buying an equity claim, and I need to know what that claim is worth.

Conservative sum-of-parts: Urban Development at 12× segment income = ¥302,220M. Media at zero (losing money). Net of net debt: minus ¥530,869M D4. Cross-holdings at book: ¥182,235M F71F72. Investment-property fair-value uplift post-tax: roughly ¥42,630M D3. Land premium above book: I'll be generous and add ¥50,000M. Conservative IV: roughly ¥46,000M — far below the current market cap of ¥569,213M D1. To get anywhere close to a fifty-cent dollar, I would need to buy at roughly ¥2,000/share, and that would still require believing in media recovery and successful debt refinancing.

Terrestrial television advertising in Japan is in structural secular decline E4. The scandal accelerated what streaming and digital were already doing to the ad market F103. The Pony Canyon anime write-downs E22 confirm that the headwinds go beyond Fuji TV's advertiser relationships. Management is pivoting — but a melting ice cube funded by new debt is not a turnaround I want to own.

On P81: the affair is described in risk disclosures as a "human-rights and compliance affair" at the subsidiary E3. The primary tail is operating — ad revenue loss already reflected in FY2026 results F78F103. No separate legal provision amount is disclosed in the evidence I have. If there are personal-harm compensation claims outstanding and unquantified, that creates an additional unbounded tail; the ledger is silent on specifics, so I mark this data-insufficient. It is a gap, not a reason alone to pass — P1 and P20 already end this.

This is a pass. The real estate is real. The Urban Development segment is genuinely good. But at 1.04× book D2 with ¥530,869M net debt D4 against near-zero operating cash flow, a broadcasting franchise in structural decline, and ¥277,577M of short-term borrowings that must be rolled, there is no floor worth buying and no price discount worth taking. Most companies studied should be a pass, and that is the system working. This one is a clear pass.

Fuji Media Holdings is two companies wearing one certificate. Under the Broadcasting Act it is a certified broadcasting holding company E2; on the balance sheet it is a Tokyo landlord. The filings split it cleanly: Media & Content — terrestrial broadcasting, streaming, film and anime — and Urban Development & Tourism — office leasing, real-estate dealing, hotels and resorts E1. In the year I am studying, those halves pulled in opposite directions. Media & Content lost ¥30,835M at the segment line F78 while Urban Development earned ¥25,185M F80, and the consolidated operating result was a loss of ¥8,766M F36. The group reported positive net income of ¥6,499M F15 only because it booked ¥50,021M of gains selling cross-shareholdings F40E13. Strip that one-off away and the operating engine ran backwards — operating cash flow was −¥341M F29. A student should sit with that fact before anything else: the "earnings" here are a securities-sale artifact, not the fruit of the business.

Now the first question I ask of any company — can I know its next ten years better than the people who already own it? — and I must answer honestly that I cannot. The variable that swings the consolidated result is terrestrial-television advertising E4, and it is falling: Fuji Television's broadcast-and-media revenue dropped 27.4% F103 and the subsidiary alone lost ¥32,515M at the operating line F84. Part of that is a self-inflicted wound — a human-rights and compliance affair at Fuji Television that materialized in the prior year and drove advertisers away E3 — and part is the slow structural tide pulling audiences from broadcast to streaming. I can price a landlord over a decade. I cannot tell you where Japanese broadcast advertising sits in ten years, still less whether a scandal-scarred network re-earns the trust of the advertisers it lost. That is the boundary of my competence, and an ability without a boundary is not an ability. This is where the study ends for most owners, and where it ends for me: the decisive earnings variable is unknowable from the filings (L1).

There is a way to make the broadcasting question stop mattering — buy the whole thing so far below the worth of its assets that the television business becomes a free option. This is the discipline I most associate with hidden-asset situations in Asia, and it is exactly what the market's stale "below book" tag on this name promised. But the promise has expired. The stamp price of ¥3,996 puts market capitalisation at ¥569,213M D1, which is 1.04 times book D2 — no longer below book at all. Re-mark the assets honestly and the discount does not reappear. Self-capital is ¥546,719M F50; the one hidden asset the filings actually quantify is the rental-property portfolio, carried at ¥410,530M but worth ¥476,114M — an unrealised gain of ¥65,584M D3E14. Add it and you reach roughly ¥612bn of adjusted worth against a ¥569,213M price D1: a single-digit discount, not the dollar-at-fifty-cents I require to be wrong and still safe. The Odaiba land sits on the books at cost — ¥388,581M of land group-wide F31 — and the folklore says its true value dwarfs that. Perhaps. But the ledger gives me no fair value for the owner-occupied land, only for the rental portfolio, and I will not buy a company on a hidden asset I cannot measure. That is not timidity; it is the rule that you must know what is in the book and not count what you cannot see (L27, L47).

The capital allocation deepens my caution rather than relieving it. Under activist pressure the company spent ¥249,045M buying back its own stock F60, of which ¥208,044M was cancelled F62, and it funded this not from surplus cash but from borrowing — short-term debt jumped to ¥277,577M F59 and total interest-bearing debt roughly tripled to ¥615,134M E11E12. Net assets fell from ¥830,023M to ¥561,467M and the equity ratio dropped from 56.8% to 37.3% F69F49F22F23. So the group now carries ¥530,869M of net debt D4 against an operating core that loses money, and it is paying ¥125 per share this year rising to a planned ¥200 F94F95 — a payout of 380.6% of depressed earnings F96. I hold one firm rule about leverage: it converts a temporary decline into a permanent loss. To bolt fresh, largely short-term debt onto a conglomerate whose profitable half is capital-hungry real estate and whose other half is bleeding is precisely the configuration I avoid, and it leaves the company dependent on its banks to roll that ¥277,577M year after year (L20).

Not everything here is a mark against management. Compensation is modest and unfussy — directors excluding audit and outside members took ¥129M in total, with no performance bonus paid in the loss year and no single director reaching ¥100M F102E18. The governance response to the affair is substantive rather than cosmetic: a majority-outside audit committee, a new Nomination and Compensation Committee chaired by an independent director E16, a Group Human Rights Committee E17. The cross-shareholding sell-down is real and measured — ¥49.4bn sold this year against a target below 15% of net assets F73E9E10. The holding company plainly receives cash from its subsidiaries F38F90. A student should note all of this, because it is the difference between a company being cheap-and-crooked and cheap-and-merely-hard. This one is not crooked. It is hard.

And hard, for me, is a verdict, not a hurdle to argue past. Neither of the two ways to own Fuji Media clears the bar. The earnings path asks me to underwrite a broadcasting recovery I cannot know (L1); the guidance to ¥40,100M of operating income F99E21 is a hope, not a floor I can defend. The asset path asks for a discount that is no longer there and rests on an Odaiba land value the filings will not give me (L18, L27). Bolt fresh leverage onto both (L20) and the margin for being wrong is gone. No price on the operating business rescues a business I cannot predict, and the asset discount that might have let me ignore that has closed. I set it down in the too-hard pile.

The business, in one breath. Fuji Media is two companies under one ticker. A terrestrial broadcaster (Media & Content, external revenue ¥350,889M F77) that lost ¥30,835M at the segment line F78, and a Tokyo real-estate/tourism operator (Urban Development & Tourism, revenue ¥193,495M F79) that earned ¥25,185M F80. Consolidated, the operating engine lost ¥8,766M F36. The reported ¥6,499M net profit F15 is not earnings — it is ¥50,021M of one-off gains on selling cross-shareholdings F40E13, net of the operating loss and tax. Hold that fixed: at the operating level, FY2026 was a loss year, and the positive bottom line is an asset sale wearing an income-statement costume.

What changed the balance sheet. Under activist pressure the company executed a ¥249,045M treasury buyback F60E11, cancelling ¥208,044M of stock F62, funded chiefly by short-term borrowing E12. Net assets fell ¥830,023M→¥561,467M F69F49; the equity ratio dropped 56.8%→37.3% F22F23; interest-bearing debt roughly tripled to ¥615,134M F58, of which ¥277,577M is short-term F59 against ¥84,265M of cash and deposits F51. Net debt is ¥530,869M D4. The "cash box awaiting a catalyst" is gone — the catalyst fired, and it was paid for with leverage. That reframes the thesis entirely: this is no longer a below-book asset pile (P/B is 1.04 D2, not sub-1) but a levered asset play with a money-losing operating core.

The NAV, built explicitly. Book equity attributable to owners is ¥546,719M F50, or ¥3,838/share on 142,445,818 shares (issued less treasury, F85−F91), which reconciles to reported BPS ¥3,838.08 F19. The one hidden asset the filings actually disclose is rental property: fair value ¥476,114M F89 vs ¥410,530M book F88, a ¥65,584M pre-tax uplift D3. Taxed at ~30% that adds ¥45,909M, or ¥322/share. The securities portfolio adds nothing beyond book — under JGAAP the ¥372,302M of investment securities F56 is already carried at fair value, with the ¥94,635M revaluation reserve F65 already inside equity, net of deferred tax. So adjusted NAV ≈ ¥4,160/share. Against the stamp ¥3,996, the stock trades at ~0.96× adjusted NAV — a ~4% discount. That is not a margin of safety; it is roughly fair value on the only marks I can compute.

What the NAV cannot tell me. The group's operating land sits at cost ¥388,581M F31 with no fair-value disclosure in the ledger. Fuji TV's Odaiba complex and the group's Tokyo land could hold large undisclosed gains — or not. I cannot mark it from these documents, so I flag it as the single largest resolvable-by-data unknown. It could add materially to NAV; I will not credit what I cannot compute.

Reverse-DCF: what the price assumes. Take the equity value ¥569,213M D1. At the current media run-rate of −¥30,835M F78, the operating businesses' going-concern value is negative, so the equity would be worth less than the ¥4,160 NAV — yet the market pays ~NAV. The price therefore embeds the FY2027 guide as substantially real: revenue ¥625,700M F98 and operating income ¥40,100M F99 — a ¥48,866M swing from this year's −¥8,766M loss F36 in a single year. Decompose it: hold Urban Development flat at ~¥25bn F80 and unallocated at −¥7bn, and Media must swing roughly ¥53bn, from −¥30,835M F78 to about +¥22bn. On the media revenue recovery implied by +13.4% group revenue F98, that is a ~70% incremental margin — implausible for broadcasting. Either the guide leans on lumpy real-estate dealing and condo sales E15, or it is optimistic. I have no guidance-accuracy history in the ledger to calibrate it (C6 is data-insufficient). The market is paying for a recovery I cannot verify.

Earnings power versus assets. Capitalize the FY2027 guided net income ¥26,100M F101 at 8% with no growth: EPV ≈ ¥326bn. Even the pre-scandal average net income ¥36,272M D5 at 8% is ¥453bn. Both sit below adjusted NAV (~¥593bn) and below the price (¥569bn D1). This is the textbook Japanese pattern (C77): assets worth more than the earnings they throw off, the gap being the prize for fixing capital allocation. The price sits above earnings power and just below asset value — the buyer is betting on the assets and paying ~full price for them.

Paid to wait? Forward DPS is ¥200 F95, a 5.0% yield at ¥3,996 — but FY2026 payout was 380.6% F96 and the business generated −¥341M operating cash flow F29. The dividend is funded by the balance sheet and asset sales, not operations, unless the guide is hit (FY2027e payout 104.5% F97). A leverage-funded 5% yield on a loss-making core is not a cushion; it is capital coming back out of a shrinking equity base.

The bet. Bull and bear both concede the real estate and the (already-marked) securities are worth roughly book-plus-a-little. They split on one proposition: does broadcasting recover toward the guide, or is −¥30.8bn F78 structural E4F103? At P/B 1.04 with adjusted NAV only ~4% above price, the asset discount is too thin to protect me if broadcasting keeps bleeding, and the upside needs a one-year swing I cannot underwrite. The governance response is genuinely good — activist-forced buyback executed E11E12, ¥48,595M of cross-holdings sold against a >¥100bn target F73E9, a real majority-independent nomination/compensation committee E16, no performance bonus paid in the loss year E18F102. But good governance on a fairly-priced, levered, loss-making asset play is a reason to watch, not to buy.

Verdict: watch. The two deciding unknowns — broadcasting's recovery and the undisclosed operating-land value — are resolvable (by the next one or two prints, and by fuller property disclosure), not structurally unknowable, so this is not too-hard. But there is no margin of safety worth acting on today: the asset discount is 4%, the bear case is ~30% down (¥2,800/share, NAV less roughly five years of capitalized media losses plus refinancing drag), and the upside is a recovery guide I cannot confirm. I would want price below ~¥3,400 (≈15% under adjusted NAV) and evidence the media loss is narrowing before the asset case alone protects the position.