Canaan's Crypto-Funded Buyback: The 19.9% Discount That Is Not a Discount

On Aug. 4, Canaan handed management a new lever: the right to fund share repurchases by selling the company's digital asset treasury. The same day, the market valued the entire company at $144.7 million. Simple addition of reported gross assets — roughly $130 million in crypto at Aug. 3 prices, plus $43.5 million in cash from March 31 — produces $173.5 million. That is $28.8 million, or 19.9%, above the intraday market capitalization. Ledger lines reveal what noise obscures: a loss-making ASIC maker has become the arbiter of whether its own equity or its own Bitcoin is the better asset. The market sees a discount. The ledger shows a liability structure and a pledge schedule the discount narrative ignores. Clarify the instrument, because the market has already muddled it. The buyback program is not new. Canaan announced it on Dec. 12, 2025, with a 12-month ceiling of $30 million for ADS or Class A ordinary share repurchases. By May 19, per first-quarter results, the company had spent roughly $2 million to repurchase about 2.8 million ADSs. The implied average execution price is about $0.71 per ADS. Simple subtraction puts nominal remaining capacity at about $28 million as of May 19. The Aug. 4 securities filing adds a new funding channel: management may sell part of the digital asset treasury — disclosed in June as 1,915 BTC and 3,952 ETH, including receivables and excluding customer deposits — and apply the proceeds to buybacks. Neither side of the trade has been disclosed. No new sale. No new repurchase. The authorization is a loaded weapon that has not been fired. Canaan is a hardware company in a hardware recession. First-quarter results recorded a $22.9 million gross loss, a $54.3 million operating loss, and a net loss of $88.7 million. Second-quarter revenue guidance of $35 million to $45 million reflected a declining order book in a market where Bitmain and MicroBT are competing for the same shrinking pool of machine demand. The treasury, meanwhile, has become the center of gravity for the entire valuation debate. The company has transformed from an ASIC vendor into a quasi-holding vehicle for crypto assets. That transformation is why the buyback funding channel matters. It converts the balance sheet into a capital-return instrument, and it does so at the exact moment the operating business cannot fund itself. That comparison — gross asset sum versus market cap — is a standard shorthand in crypto equity analysis. It is also a dangerous one. The shorthand assumes assets are real, unencumbered, and immediately convertible. Each assumption must be tested against the footnotes before the discount means anything. Begin with the holdings. June 30: 1,915 BTC and 3,952 ETH. At the implied valuation embedded in the $130 million estimate, that is roughly $60,000 per BTC and $3,300 per ETH — a defensible mid-summer mark. Headline holdings are not available liquidity, however. In March, Canaan had pledged 905 BTC for secured term loans and placed 100 BTC in a fixed-term product. That is 1,005 BTC encumbered before any buyback funding is extracted — over 52% of the June BTC position. When I standardized due diligence frameworks after the 2022 collapse, I wrote one rule into every checklist: a pledged asset is not a treasury asset; it is collateral with a claim attached. Liquidity is the current of truth. Canaan's headline crypto reserve has a current that is half-blocked. Cash tells the same story in a different denominator. March 31: $43.5 million in cash against $106.4 million in current liabilities. At year-end 2025, cash was $80.8 million. The company burned more than $37 million of cash in three months while carrying a liability stack more than double its cash balance. April brought roughly $42 million in customer-receivable collections — a partial bridge, not a structural one. The Q1 figures explain the burn: an $88.7 million net loss in a quarter that ended with a growing crypto pile and a shrinking order book. A buyback funded by treasury liquidation is not returning capital to shareholders. It is converting the emergency reserve into buy volume in the company's own stock. Now apply the discount. The 19.9% gap between gross assets and market cap is the number the headlines carried. The arithmetic is real, but it omits the right-hand side of the balance sheet. Subtract the $106.4 million of current liabilities from the $173.5 million gross sum and the implied net position drops to roughly $67 million — below the company's own market capitalization. That is not a same-date net asset value; the crypto estimate is Aug. 3, the cash is March 31, and the liabilities are March 31. It is, however, a directional correction. The 'discount' is a gross-asset discount, not a net-asset discount. A buyer of every share inherits the liabilities, not just the crypto. My 2020 DeFi work taught me to measure the truly available liquidity before counting a yield. The same discipline applies here: the unencumbered treasury — not the pledged one — is the only portion that can fund a buyback without adding leverage. The disclosed execution data is thin but telling. $2 million spread over 2.8 million ADSs is an average price of roughly $0.71. Canaan's ADSs trade in dollar-and-below territory, a region institutional buyers treat as delisting risk rather than deep value. That the board authorized, and management partially executed, purchases at that level contains an implied admission: management considers its own equity the most undervalued asset on the list. That is a brutal statement in a bull market because the same management has chosen to accumulate Bitcoin and Ethereum quarter after quarter. The inconsistency between accumulating BTC and liquidating it to buy equity is the signal. It means the equity is not the preferred asset — operating liquidity is. Every buyback executed with BTC proceeds is a forced-choice trade: the company needs buy volume or working capital, so it sacrifices the asset that anchors its valuation narrative. The June operating update specifies that the disclosed 1,915 BTC and 3,952 ETH are 'including receivables and excluding customer deposits.' That phrasing is itself a forensic clue. Miners often hold customer deposits in digital assets — prepayments for machine orders that are not the company's unencumbered property. Excluding them tightens the asset base. Including receivables widens it. The precise mix matters because customer-deposited BTC carries an obligation to deliver machines or refund value. If a share of the disclosed BTC is, in substance, a liability rather than an asset, the $130 million figure the market anchors on is already too generous. My experience auditing protocol claims taught me to treat 'treasury' as a hypothesis, not a fact, until the encumbrances are disclosed. The program's capacity is small relative to the treasury but large relative to cash. $28 million of nominal remaining authorization is about 22% of the $130 million treasury and 64% of the March cash balance. In the most bullish scenario, management sells $28 million of BTC and repurchases roughly 39 million ADSs in the $0.71 region. Per-share metrics improve mechanically. The arithmetic is not the problem. The sequencing is. Treasury sale and share repurchase are two market events separated by days or weeks. The sale moves the BTC price. The purchase moves the ADS price. The net effect on shareholder value depends on which market absorbs the flow with less slippage — a microstructure question the SEC filing does not answer. Efficiency is the only permanent alpha. An undisclosed execution plan is not an efficiency signal. The bullish read says: Canaan's management is willing to tap a $130 million war chest to defend the share price. That is causation dressed as correlation. The forensic read: Canaan has $43.5 million of cash against $106.4 million of current liabilities, a quarterly net loss of $88.7 million, and a hardware cycle the company itself calls weak. The treasury is the only unencumbered asset large enough to matter. Authorizing it as a buyback funding source transfers optionality from the balance sheet to the trading desk. It buys time, not solvency. If fully executed, the company converts $28 million of BTC into stock held by the company — an asset that produces no revenue, no collateral value, and no protection against the next ASIC order cancellation. The market will call it capital return. The ledger calls it asset substitution. Strategy's May decision to list Bitcoin sales as a repurchase source offers the same playbook, larger balance sheet, different risk profile. Canaan is running the leveraged version with a negative operating margin underneath. The pledge schedule is the detail most coverage omits. 905 BTC were pledged for secured term loans at March 31. The June operating update did not disclose whether those pledges were released. If they remain, their sale requires lender consent or repayment of the associated loans — in cash, or through delivery of the BTC itself. That means the 'funding channel' the market celebrated on Aug. 4 may apply only to the unencumbered remainder: roughly 810 BTC plus the 3,952 ETH, after excluding the 100 BTC in the fixed-term product. The available pool is not $130 million. It is a subset. In 2018, when I audited shielded transactions, I learned that a proof can be valid and still useless if the assumptions are violated. The valid assumption here is that Canaan holds BTC. The violated assumption is that it can deploy that BTC without tripping over a lender's security interest. Code does not lie, only developers do — and balance sheets carry liens. There is also the opportunity-cost problem. A bull market rewards patience with holdings and punishes forced liquidity. If Bitcoin rises 30% over the next twelve months, every BTC sold to fund a buyback at a gross-asset discount of 19.9% is a double surrender: the company sells an asset that would have appreciated to buy equity in a business that lost $88.7 million in a single quarter. The share buyback only outperforms the BTC sale if the ADS appreciates more than the BTC would have. Given the operating losses, that is a bet on market sentiment, not mathematics. I have written this before, and I will write it again: the graph clarifies what sentiment confuses. The graph of Canaan's cash will show a declining line. If this funding channel is used, the graph of its BTC will also decline. Two declining lines are the actual signal, and neither supports the buyback narrative. CoinShares' March 2026 mining report drew a clean line across the sector: stressed miners are selling coins, stronger operators are pivoting to AI, and listed mining stocks are becoming less pure Bitcoin proxies. Canaan sits at the first node. Q1 produced a gross loss, meaning every machine sold in the quarter destroyed gross profit. The AI pivot never materialized. The buyback channel announced on Aug. 4 is therefore not a capital-return innovation — it is the natural endpoint of a company whose core product no longer generates cash. The market treated the announcement as a positive surprise because the headline read 'management will defend the stock.' The ledger read it as 'the last unencumbered asset gets a sales process attached.' In a bull market, that inversion is exactly where excess insight lives: the spread between the headline and the balance sheet. Canaan's own statement that 'working-capital needs and board oversight set the limit' is the sentence I would underline. It concedes the risk: the limit is not the discount, not the float, not the market cap. It is the board's assessment of how much cash the hardware business needs to survive. A buyback that drains working capital to defend per-share metrics is precisely the decision that post-mortems later flag as a governance failure. The board is the correct gatekeeper. What the market cannot verify from the Aug. 4 filing is whether the board has defined a floor. That uncertainty is the real residual risk. The next quarterly report will settle the question. Two numbers, on the same page: the BTC/ETH balance and the cumulative repurchase count. If the treasury declines and the buyback count rises, management is executing a liquidation-style capital return that will end the day the program's ceiling is reached. If the treasury holds and the buyback total stays near $2 million, the Aug. 4 filing was a signal with no follow-through — the bear-market habit of buying optionality at the cheapest possible price. Canaan's real trade is not the buyback. It is the conversion of a mining business into a treasury vehicle with a liquidation schedule. Bear markets demand disciplined forensics; bull markets demand the same discipline with louder noise around it. The buyer of Canaan ADSs is placing a leveraged bet that management's incentive to hold Bitcoin exceeds its incentive to defend the share price. The ledger says the incentive has two sides. The arbitrage between them is the only thing this financing channel will ultimately reveal.