GUMI INC

Company Overview

GUMI INC is a Japanese developer and operator of mobile games that has increasingly become a hybrid of a licensed-IP game studio, a Web3/blockchain operator, and a corporate holder of digital assets. The distinction matters: the quoted equity is not simply a game-company claim on operating cash flows. A large part of reported net assets is exposed to the liquidity and prices of tokens created or accumulated through its Web3 activities.

The latest official disclosure is the unaudited FY2027 first-quarter earnings release for the three months ended 31 July 2026, released 11 September 2026. It supersedes the audited FY2026 annual base for near-term diagnosis. The latest clean official annual base is FY2026, ended 30 April 2026 and filed 29 July 2026. More recent data is partial and unaudited; the first-quarter release contains no cash-flow statement and management provides no FY2027 earnings guidance because it considers its operating environment too volatile.

Current market data supplied is a share price of ¥202 and market capitalization of ¥10.887 billion as of 14 September 2026. At 31 July 2026, the company reported cash of ¥3.058 billion and identified interest-bearing borrowings and bonds of ¥4.695 billion, or net debt of ¥1.637 billion. This is a balance-sheet calculation, not a statement about total liquidity: it excludes the company’s digital-asset holdings because their realizable value is now the central uncertainty.

The clean annual earnings base was weak at the operating level. FY2026 audited annual data showed revenue of ¥9.184 billion, operating income of ¥83 million, ordinary income of ¥2.170 billion, and attributable net income of ¥1.455 billion. The large gap between operating and below-the-line profit reflects investment and digital-asset-related items rather than a proven recurring operating franchise. In the latest unaudited Q1 FY2027 data, revenue was ¥1.628 billion, operating loss ¥186 million, ordinary loss ¥2.018 billion, and net loss ¥1.869 billion.

How the Company Makes Money

GUMI reports two businesses. The renamed Neo-Media Entertainment segment develops and operates mobile games, principally using third-party intellectual property. It earns game revenue from players through Apple and Google app stores, alongside development commissions and potentially royalties from participation in content-production committees. The stated strategy is to reduce original-title risk by combining its game engine and development capabilities with established external IP, then to supplement this with commissioned development, console/Steam content, and adjacent entertainment activities.

The Neo-Crypto segment includes blockchain-game and platform activity, validator-node operations, staking rewards, digital-asset management, systems for crypto-asset accounting, fund investments, and direct holdings of tokens. Its reported revenue can include token receipts; its economic result also depends heavily on token prices, liquidity, and accounting treatment. This segment generated Q1 revenue of ¥516 million and operating income of ¥312 million, versus the game segment’s ¥1.112 billion of revenue and ¥498 million operating loss. The crypto business was therefore the sole positive operating segment in the latest quarter, but this does not establish that it is durable or cash-generative through a full cycle.

The company is not physically capital-intensive, but it is economically capital-intensive in a less attractive way: game development, advertising, IP acquisition, token accumulation, and venture/fund investments can consume cash before outcomes are known. FY2026 operating cash flow was negative ¥1.398 billion and investing cash flow negative ¥3.589 billion. The prior four fiscal years also recorded negative operating cash flow. This is not the profile of a demonstrated owner-earnings compounder.

There are useful capabilities here: game-development experience, relationships with IP owners, and early Web3 operational knowledge. They are capabilities, not a strong moat. Licensed IP belongs to licensors, platform distribution belongs to Apple and Google, players can migrate quickly, and nodes or token-management services can be replicated by capable, funded competitors. GUMI may have a franchise opportunity around individual titles or relationships, but the evidence does not support treating it as a durable high-return compounder.

Why the Stock Fell

No verified historical price series was supplied, so the percentage decline and the exact contribution of each event cannot be established here. The observable recent facts nevertheless identify a coherent market narrative: weak game economics, substantial token-mark-to-market losses, and a challenge to the liquidity of the digital assets supporting reported book value.

• 11 September 2026, Q1 FY2027 results: revenue rose 20.2% year on year, but operating profit reversed from a ¥73 million profit to a ¥186 million loss. The newly released “JoJo’s Bizarre Adventure OraOra Overdrive” generated revenue below management’s expectation, while development spending ahead of the August launch of “My Hero Academia UNITED SURVIVAL” increased costs. The game segment’s operating loss widened to ¥498 million from ¥190 million a year earlier.
• 11 September 2026, token-related loss recognition: the same quarter included a ¥1.538 billion digital-asset valuation loss, a ¥153 million digital-asset disposal loss, and a ¥188 million equity-method investment loss. These items drove the ¥2.018 billion ordinary loss and ¥1.869 billion net loss. They are not merely a change in sentiment; they reduce reported equity unless subsequently reversed.
• 14 August 2026, FCT delisting from MEXC: the company disclosed that FCT was delisted from MEXC. At 31 July, FCT represented ¥7.987 billion of the ¥13.301 billion digital-asset balance.
• 4 and 9 September 2026, OSHI delisting announcements: BITPOINT announced an OSHI delisting effective 28 October; Gate.io announced a delisting effective 16 September; and MEXC was described as highly likely to delist OSHI on 12 September. OSHI carried a ¥2.656 billion book value at 31 July.

The dominant inference is that the market stopped viewing the reported token balance as a near-cash asset and began treating the company as a cash-burning game developer with a concentrated, potentially illiquid digital-asset portfolio. That inference is economically sound. It would be an overreach, however, to call every token balance worthless: management stated that FCT and OSHI continued to trade on other exchanges at the time of disclosure. The issue is that tradability, depth, and disposal value are unverified at the required scale.

What the Market Is Assuming

The market capitalization of ¥10.887 billion is below the FY2026 reported net assets of ¥22.134 billion and below the supplied FY2026 NCAV calculation of ¥13.756 billion. On their face, those figures look like a classic asset discount. The market is instead implicitly assigning a severe haircut to the assets and little value to the reported earnings stream.

One-time or cyclical factors. The Q1 loss contains discrete mark-to-market and disposal losses on digital assets. Those losses can reverse if liquid markets and prices recover. Likewise, pre-launch development and user-acquisition spending can be temporary. Yet neither should be casually normalized away: the company’s FY2026 operating income was only ¥83 million, and the Q1 game-segment loss worsened despite revenue growth. A recovery in token prices would repair accounting equity, but it would not by itself prove a better game business.

Medium-term operating headwinds. Revenue has contracted sharply over the available multi-year record: ¥18.942 billion in FY2022, ¥16.010 billion in FY2023, ¥12.067 billion in FY2024, ¥8.943 billion in FY2025, and ¥9.184 billion in FY2026. The supplied longer-period revenue CAGR is negative 9.5%. Operating income over the same years was negative ¥2.262 billion, positive ¥447 million, negative ¥5.040 billion, positive ¥371 million, and positive ¥83 million. These are not the economics of a stable recurring-content business. FY2026 operating cash flow was negative ¥1.398 billion after negative operating cash flow in each of the prior four reported fiscal years.

Long-term threats. The market appears to assume that licensed game titles may remain hit-driven and that token-based assets may be structurally discounted because liquidity can disappear when exchanges delist them. The evidence supports both concerns. It also must price dilution: the supplied share-count change is +8.9% over FY2026, and shares outstanding increased from roughly 49.5 million at FY2025 year-end to 53.9 million by FY2026 and remained 53.9 million in Q1 FY2027. FY2026 stock-based compensation expense was ¥397 million; it is a real shareholder cost and is retained in the normalized earnings estimates below.

There is an important evidence gap. The disclosures do not provide independently verifiable current market depth, disposal proceeds, or current prices for the concentrated FCT and OSHI positions. Nor do they disclose title-level lifetime customer-acquisition economics, player retention, or a recurring-revenue split sufficient to underwrite a durable game cash-flow base. A quoted discount to book value therefore cannot be converted into a margin of safety without resolving asset realizability.

Temporary or Structural?

Game-title underperformance: structural but survivable. The damaged mechanism is the conversion of development expense and marketing into retained, paying players. “JoJo” revenue fell short of expectation, while the segment continued to lose money. A failed title is temporary; the underlying hit-driven, externally licensed-IP model is structural. It can improve within three years through a successful title, lower fixed cost, or more commission work, but there is no evidence that a repeatable, high-return release process has been established.

Token concentration and exchange delistings: structural but survivable at the corporate level, potentially permanent for asset value. The damaged mechanism is asset liquidity and mark-to-market value. A ¥10.643 billion combined carrying value for FCT and OSHI represented about 80% of Q1 digital assets. If the tokens lack deep, functioning markets, the economic impairment can be irreversible even if accounting recognition is deferred. The company can survive a large write-down because its liabilities were ¥6.288 billion against reported Q1 assets of ¥26.617 billion, but survival is not the same as preservation of per-share value. Recovery inside three years is possible, not underwritable.

Dependence on platform owners, licensors, and user acquisition: structural but survivable. Apple and Google collected or processed 52.6% of FY2026 revenue, while third-party IP determines access to recognizable content. The company can diversify titles and channels, but it does not control these gates. The mature mobile-game market also raises development and marketing costs. This creates a structurally weak bargaining position rather than a fatal flaw.

Dilution and capital allocation: structural damage to per-share compounding. Repeated issuance can fund projects and extend the operating runway, but it transfers the economics of uncertain investments to existing shareholders. This is reversible within three years only if the company generates genuine free cash flow and retires shares or stops issuing equity. Neither outcome can be assumed from the current record.

Time-as-a-moat test: a well-funded competitor could build a credible game studio, secure licensed IP, and run validator nodes in two to five years. In ten years, it could replicate virtually all of the operating model. What cannot be cheaply rebuilt is a specific hit title, a particular relationship, or a liquid token ecosystem—but those are contingent assets, not durable company-owned moats. The verdict is therefore ESSENCE for the quality of earnings and asset reliability, with a TIME component in the Q1 marketing cycle and token-market mark-to-market loss.

Is the Market Wrong? By How Much?

Moat-and-mispricing score: 3/10. The low score does not mean the shares cannot rise sharply. It means the apparent statistical cheapness is mainly a claim on uncertain assets, while the recurring operating engine has not earned a low required return. The market may be right to reject reported book value as intrinsic value. It is not demonstrably wrong at ¥202; the current value is close to a conservative base case and remains highly sensitive to token realizability.

The supplied FY2026 “worth-alive” owner earnings estimate is ¥2.011 billion, implying an 18.47% yield on the 14 September market capitalization. This is a model estimate, not audited earnings, and its supplied free-cash-flow conversion is negative 100.53%. It cannot be capitalized at face value because FY2026 reported operating income was only ¥83 million, operating cash flow was negative, and the following quarter produced an operating loss. The valuation therefore excludes token valuation gains and losses from normalized owner earnings and values the token/investment portfolio separately at explicit haircuts.

The valuation is FY2026-based and adjusted with the unaudited Q1 FY2027 balance sheet and events through 14 September 2026. The current 53.901 million shares outstanding are used for per-share values. These are intrinsic-value estimates, not price targets.

The asset haircuts are the valuation’s decisive assumption. The bear case values FCT and OSHI at zero, other digital assets at 50% of carrying value, marketable securities at 50%, and related-company securities at 25%; it assigns no value to other investment assets. The base case values FCT at 30% of carrying value, OSHI at zero, other digital assets at 75%, marketable securities at 70%, related-company securities at 50%, and other investment assets at 50%. The bull case assumes FCT retains 70% of carrying value, OSHI 25%, other digital assets 100%, and higher recoveries for investments. These are my estimates, not company disclosures or market quotes.

The base case deliberately assumes only ¥300 million of normalized annual owner earnings after stock-based compensation, rather than accepting the supplied ¥2.011 billion model figure. That amount is above FY2026 operating income but below what a sustainable high-return franchise would earn; it requires the game losses to narrow and recurring crypto/node or commission income to remain positive. The bull case requires both significant asset recoverability and a repeatable operating recovery. The bear case illustrates why NCAV is not a floor when a major part of current assets may be illiquid tokens.

At the supplied current market capitalization, the market appears to be assuming more than a liquidation discount. Using the base asset haircuts, it implicitly assigns roughly ¥4.5 billion to operations after net debt, equivalent to about ¥675 million of annual owner earnings at a 15% required yield. That is far above FY2026 operating income and inconsistent with Q1’s operating loss unless the latest game spending is genuinely temporary and the crypto segment’s operating contribution is recurring. The market is therefore not obviously too pessimistic; it is already underwriting a meaningful recovery.

Key Facts, Estimates, and Judgments