At 2:55 p.m. EDT on Aug. 4, the market priced Canaan at $144.7 million. Canaan's balance sheet told a different story: roughly $130 million in digital assets plus $43.5 million in cash. Gross sum: $173.5 million.
The gap: $28.8 million. Nineteen point nine percent. The company traded below the value of the assets it already held. On the same day, Canaan filed with the SEC that management can now sell the crypto treasury to fund share buybacks. The last large flexible asset becomes fuel for stock repurchases.
Two possible endings. The lever closes the mispricing, or it accelerates the burn. I've watched treasury-funded buyback structures across a decade of crypto capital markets. They rarely end where the narrative begins.
The mechanism is simple. The Aug. 4 filing lets management use digital asset proceeds under an existing buyback program. That program began Dec. 12, 2025, with a 12-month ceiling of $30 million for repurchases of ADSs or Class A ordinary shares. By May 19, Canaan had spent roughly $2 million to buy back about 2.8 million ADSs. Simple subtraction leaves roughly $28 million in nominal capacity. The August release doesn't update the total, so the true unused authorization remains unknown. Execution stays undisclosed on both sides: no announced treasury sale, no announced repurchase.
Now look at the company behind the lever. Q1 2026: $22.9 million gross loss. $54.3 million operating loss. $88.7 million net loss. Q2 revenue guidance of $35–45 million was in line with a downgraded forecast, not with health. ASIC demand collapsed, and the hardware business bleeds at the gross margin level.
Cash fell from $80.8 million at year-end to $43.5 million on March 31. Current liabilities: $106.4 million. The buffer sits in negative working capital territory. April brought about $42 million of customer-receivable collections, but the structural picture remains tense. The June operating update adds the key number: 1,915 BTC and 3,952 ETH on the balance sheet, including receivables and excluding customer deposits. That's the largest pool of value the company controls. It's now the designated buyback account.
One calibration note: the $173.5 million figure uses Aug. 3 crypto prices, a March 31 cash balance, and an Aug. 4 intraday market cap. The dates don't align, and the calculation omits liabilities and asset restrictions. Directionally, though, the picture holds. That's exactly why the disclosure matters.
Start with the encumbrances. 905 BTC sit pledged for secured term loans. Another 100 BTC are locked in a fixed-term product. Against the 1,915 BTC balance, that's 47.3% pledged and 52.5% encumbered or locked. The free bitcoin float is smaller than the headline suggests. The Aug. 4 filing doesn't state whether those restrictions changed. I treat headline treasury numbers as theoretical until the collateral agreements are updated. Empirical verification starts with reading the pledges, not the press release.
Now run the full balance sheet. The $173.5 million gross sum sits 19.9% above the market cap. But $106.4 million in current liabilities sits on the other side of the ledger. Strip those out and the equity cushion is far thinner than the discount implies. The market isn't ignoring a risk-free arbitrage. It's pricing the haircut on a gross asset pile that must clear debt first. That's the difference between headline value and net realizable value.
Here's what the persistent discount actually says. If the market believed the treasury was fully liquid and unencumbered, the market cap would sit closer to gross asset value. The gap has persisted through this announcement and will likely persist after it. That persistence is smart money's estimate of the liquidation haircut: pledged coins, lockups, liabilities, and the operational burn all feed into the spread. Discount to book is not a defect in the market. It's a verdict.
Arbitrage is just efficiency with a heartbeat. In this case, the heartbeat is a company losing $88.7 million per quarter.
Then follow the circular logic. Sell BTC to buy shares. Each repurchase shrinks the float and mechanically supports per-share value. Each sale shrinks the treasury—the primary rationale for the stock's floor. Both sides of the ledger move in opposite directions. This isn't value creation. It's a transfer with slippage attached.
Timing matters. The authorization existed since December. By May 19, deployment sat at just $2 million of $30 million. Then, in August, the treasury is formally designated as buyback fuel. What changed? The cash position hasn't recovered. April's receivable collections helped, but $43.5 million against $106.4 million in current liabilities leaves no discretionary cash. The crypto is the only flexible asset large enough to move the share price. The filing isn't about confidence. It's about access.
Collateral risk compounds the problem. 905 BTC are pledged. If bitcoin falls, loan-to-value ratios deteriorate. Forced liquidations cascade into the same market where the buyback sits. I spent 72 hours tracing the Luna collapse in 2022, following the oracle failure into the death spiral. The lesson stuck: when everyone assumes a large asset pool is liquid, the first margin call reveals how fragile that assumption was. A treasury serving as both collateral and buyback fuel cannot satisfy both obligations at once.
The peer comparison sharpens the picture. Strategy floated a similar idea—selling Bitcoin to fund repurchases. Headlines followed. But Strategy's balance sheet can absorb a Bitcoin drawdown without existential stress. Canaan cannot. Its crypto treasury is roughly 90% of its market cap. Selling it in size to repurchase shares at a discount signals defense, not conviction.
Execution drag completes the analysis. A $28 million BTC liquidation moved through the open market shakes the tape. On-chain observers track known miner wallets. Front-running bots monitor the same mempool the treasury desk would use. Slippage on a single-day sale of that size can easily hit 1–2%, and the signaling effect—a listed miner selling its stack—pushes price action beyond the raw volume. The filing describes the plan. The tape reveals the cost.
Retail interpretation: management is backing the stock with real assets. Bullish. The forensic read: a cash-burning manufacturer just unlocked the keys to its last liquid reserve to defend a share price. That's not conviction. That's damage control.
The structural trap is circularity. The market gives Canaan a floor because of the treasury. If that treasury is consumed to buy shares, the floor descends. For the discount to close, the buyback must lift the stock faster than the treasury depletion pulls valuation down. With $54.3 million in quarterly operating losses, the arithmetic doesn't close in the buyback's favor.
Patterns across the mining sector confirm the direction. CoinShares' latest report shows stressed operators selling coins, stronger firms borrowing to pivot into AI, and listed miners becoming less pure Bitcoin proxies. Canaan sits in the stressed column. A buyback is financial engineering layered on a hardware cycle with no visible bottom. The market isn't confused about that. It's pricing it.
You don't trade the narrative. You trade the liquidation path. This path runs from a filing through a sell order into a market already watching miner flows. There's also the information asymmetry: management hasn't disclosed quantity, timing, or price levels for either leg. They can sell crypto into strength and buy stock into weakness while investors watch lagging filings. The announcement doesn't align incentives. It widens the gap between insiders and everyone else.
ZK proofs don't make a balance sheet liquid. Revenue does. Canaan's revenue is contracting, gross margin is negative, and the cash runway depends on an asset it just volunteered to spend. Code is law, but gas fees are the reality. The plan exists in a filing. Execution happens in a live market where every BTC sale leaves an on-chain footprint.
Track the buyback burn rate. If Canaan deploys the remaining ~$28 million quickly, watch Canaan-linked wallets for BTC and ETH outflows. The treasury-to-market-cap gap is the scoreboard. If it narrows toward 5%, the discount trade is done. If it widens past 25%, the market is telling you the treasury is worth less than marked.
Ask which side moved when the gap closes. A shrinking market cap is not convergence. It's retreat. The real question isn't whether they can buy back the stock. It's whether they can buy back trust in a business that loses $88.7 million a quarter. Treasury arithmetic is easy. Solvency is a different equation.
