Zhihu (ZH): RMB 3.4 Billion in Cash on the Balance Sheet, Market Cap at Half That — Why I'm Still Not Buying
When the discount never closes: governance structure, management quality, and cash burn rate as the real variables in a deep-value trap.
The numbers are seductive
As of December 31, 2025, Zhihu had RMB 3.369 billion in cash and equivalents, plus RMB 240 million in term deposits, RMB 841 million in short-term investments, and RMB 158 million in long-term investments. Subtract RMB 72 million in borrowings and leases and RMB 71 million in noncontrolling interests, and you get net financial assets of roughly RMB 4.47 billion. Every line item was cross-checked against the original SEC filings.
The company’s market cap at USD 2.95/ADS and 88.15 million ADS outstanding: roughly RMB 1.82 billion.
You’re buying a pile of cash and wealth management products at less than forty cents on the dollar. Sounds like free money.
Before you reach for your wallet, three questions need answers: how fast is this cash shrinking, whether you have any mechanism to claim it, and whether the person controlling it deserves your trust. The answers are not encouraging.
The ice cube is melting
Zhihu’s operating business does not generate cash. Full-year 2025 operating cash flow was negative RMB 364 million; adjusted operating loss was RMB 269 million. The barely positive adjusted net income of RMB 37.9 million was propped up by RMB 232 million in investment income — mostly an unrealized paper gain from a 2021 pre-IPO bet on 58 Daojia, a home services platform with zero strategic relevance to Zhihu’s core business.
The largest revenue line, paid membership, is shrinking: average monthly subscribers fell from 15.0 million in 2024 to 13.5 million in 2025, then to 12.2 million in Q4. The second-largest line, marketing services, has declined for three consecutive years. The newest narrative — AI-native search and GEM (Generative Engine Marketing) — has zero separately disclosed revenue to date.
The company’s own reported liquid asset total tells the story: RMB 5.463 billion at end of 2023 → RMB 4.859 billion at end of 2024 → RMB 4.451 billion at end of 2025. That’s RMB 400–600 million evaporating every year. This is not a cash-generating asset. It is a melting ice cube.
The good news: the melt rate is decelerating — OCF losses narrowed from RMB 790 million in 2023 to RMB 364 million in 2025. Extrapolate the trend and OCF could cross zero around 2027. But trend extrapolation and management execution are very different things.
The case against management, in five layers
One: strategic drift. Q&A community → knowledge payments → paid membership → vocational education (impaired and written down) → web fiction / short dramas → AI search / GEM / data services. Each pivot chased the zeitgeist. None reached category leadership before the next narrative began.
Two: the business cannot sustain itself. S&M, R&D, G&A, and headcount have all been cut. OCF is still negative. Gross margin at 59.9% looks fine, but every RMB 100 in revenue leaves only RMB 60 in gross profit — of which RMB 46 goes to sales and marketing and RMB 19 to R&D, before even touching fixed platform costs.
Three: poor capital allocation. RMB 300 million went into a pre-IPO bet on 58 Daojia — a home services marketplace with zero synergy to a knowledge community. The IPO was blocked by regulators and remains incomplete. RMB 55 million went into an AR smart glasses startup. RMB 30 million went into a Sichuan-based government-guided fund as an LP commitment — a textbook local government relationship cost, not a market-driven investment. Idle cash chased whatever was fashionable.
Four: buybacks are offset by stock-based compensation. RMB 937 million spent on repurchases over 2023–2025. But SEC 13D/A filings reveal that a material portion of repurchased shares ended up in a trust under the 2022 share incentive plan — holding 10% of Class A shares as of November 2025, earmarked for future employee grants. This is not capital return. This is buying back shares with shareholder money and handing them to insiders.
Five: opaque disclosure. The company reports as a single operating segment. No segment-level margins. Does the serialized fiction business (Yanxuan Stories) make money? What are the IP licensing revenue-sharing terms? Does the AI data business have any signed contracts? All unknown. You can either take management’s narrative on faith or not, and their track record does not support faith.
Governance locks you out
Founder and CEO Zhou Yuan holds all Class B shares (10 votes each) through MO Holding Ltd, commanding approximately 43.6% of total voting power — on just 14% economic ownership.
No outside shareholder can win a proxy fight. Zhou Yuan has a unilateral veto over any special resolution. You cannot force a dividend, block a wasteful investment, or remove him from the board.
The classic deep-value catalyst — “the stock gets cheap enough to attract an activist” — does not function under a dual-class share structure. You can accumulate all the Class A shares you want; they will never outvote the Class B super-voting bloc.
Your investment in this stock is 100% dependent on Zhou Yuan’s goodwill. The only realistic paths to breaking the deadlock are narrow and unpredictable: a voluntary take-private by Zhou Yuan himself, the triggering of the Hong Kong Stock Exchange’s WVR sunset clause (which requires his departure or death), or a Chinese regulatory intervention. None of these can be initiated or timed by an outside investor.
Pricing a melting ice cube you can’t touch
With no catalyst to wait for and no governance lever to pull, the only investable thesis reduces to this: buy at a price where, even if management continues to destroy value for N more years, you roughly break even on the residual assets — and cap your position size so the downside is immaterial to your portfolio.
Starting assets on a conservative basis: cash RMB 3.610 billion + short-term investments at a 40% haircut (RMB 504 million) − liabilities RMB 143 million = RMB 3.971 billion. Long-term investments (58 Daojia, INMO, AEZ fund, Tianfu fund) are written to zero.
Annual cash burn of RMB 500 million — not a guess, but the two-year average of actual liquid asset decline (RMB 604 million in 2024, RMB 408 million in 2025), with both OCF losses and buyback spending narrowing.
Denominator: 100 million fully diluted ADS. Exchange rate: 6.99.
Three-year burn, no governance discount → (3,971 − 1,500) ÷ 100 ÷ 6.99 = USD 3.53
Three-year burn, 30% governance discount → (3,971 − 1,500) × 0.7 ÷ 100 ÷ 6.99 = USD 2.47
Five-year burn, no governance discount → (3,971 − 2,500) ÷ 100 ÷ 6.99 = USD 2.10
Five-year burn, 30% governance discount → (3,971 − 2,500) × 0.7 ÷ 100 ÷ 6.99 = USD 1.47
The 30% governance discount is not a VIE discount — even if you fully trust the VIE contractual pass-through, you should still demand compensation for being a minority shareholder with zero influence over a founder who controls 44% of votes on 14% economics.
Conclusion
At USD 2.95, you are already below the three-year-burn floor of USD 3.53. The market is pricing in more pessimism than “management burns cash for three years and you get the residual” — it is implicitly layering in governance risk, VIE risk, or a belief that the burn rate will accelerate. If you think three years of continued burn at current rates is the right base case and the VIE channel is functional, the current price is already offering you roughly 20% of margin below the floor. That is enough to nibble.
A small position at current levels is defensible. Not because the management is good — they aren’t — but because the price is doing the work that management won’t. Even on conservative assumptions (short-term investments at a 40% haircut, long-term investments at zero, fully diluted share count), you are buying a dollar’s worth of liquid assets for roughly sixty cents. If OCF inflects toward zero over the next two to three years — which the trend line suggests is plausible even without management heroics — the stock re-rates to USD 6–7 as the market swaps its “melting ice cube” narrative for an asset-value framework.
USD 2.0–2.5 is where you add more aggressively — that zone reflects three to five years of burn plus a 30% governance discount, and gives you a margin of safety against the tail risk of a sudden large capital misallocation. At those levels, the math works even if management keeps chasing trends and keeps diluting you with SBC.
Position size: no more than 2% of portfolio. Not because the asset isn’t cheap, but because you have no tool to convert “cheap” into “realized.” Graham-era net-nets worked because one-share-one-vote governance allowed shareholders to apply pressure. Under Zhihu’s dual-class structure, that mechanism does not exist. You are making a bet on arithmetic, not on agency.
Cheap is not a catalyst. The mechanism that converts cheap into realized value is the catalyst. Until you see that mechanism, position sizing is your only risk control.
Disclaimer: This article is for informational and educational purposes only and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any security. The author may or may not hold positions in the securities discussed. Investing involves risk, including the potential loss of principal. Readers should conduct their own due diligence and consult a qualified financial advisor before making any investment decisions.