Canaan's 1,915 BTC Looks Like a Treasury. Most of It Was Collateral.

Canaan reported $31.9 million in second-quarter 2026 revenue on Sept. 8, missing the $35 million to $45 million range it guided in May. Product revenue fell to $13.6 million from $42.9 million in the first quarter. Management then guided third-quarter revenue down to $11 million to $15 million. Those are the numbers most desks will quote. The number that actually matters sits two paragraphs deeper in the filing. Of Canaan's 1,915.5 BTC at June 30, 1,117 BTC was pledged as collateral against secured term loans, and 100 BTC was allocated to a fixed-term product. Those coins were booked as cryptocurrency receivables worth $70.9 million — a separate line from the $47 million classified as cryptocurrency assets. Two line items, two different kinds of claim, one headline balance that most coverage will flatten into a single treasury figure. To see why composition matters, hold Canaan's two businesses apart. It manufactures ASICs, and it mines with them. Manufacturing is a leveraged bet on capex cycles and hashprice. Mining is an operating business that converts electricity into Bitcoin at a margin. For most of 2025, the equity market rewarded a third, newer thing: the treasury. Post-ETF, listed crypto-adjacent firms discovered that coins on the balance sheet produced multiples that operating income could not. Canaan is not alone in that cohort, and that is the problem. The treasury trade worked while the coins were unencumbered and the market paid a premium for holding them. The premium is the part that does not survive a bear market. By the second quarter, that tension had resolved into arithmetic. Mining produced 243 BTC and generated $17.7 million in revenue. Management described the segment as cash-contributive before depreciation, which is a narrower claim than profitability and, to their credit, an honest one. The equipment segment lost its pricing power. Lower computing power sold, lower average selling prices, and a 68% sequential revenue decline are what a hashprice squeeze looks like from the seller's side of the table. I spent 2022 verifying fraud-proof and validity-proof code line by line instead of watching my portfolio, which lost 80% anyway. The lesson that survived was unglamorous: a balance-sheet line item is a claim, not a coin. History rhymes, but the code doesn't — and neither does an accounting footnote. Receivables are a promise owed to Canaan; crypto assets are coins Canaan controls. Lenders holding the pledged BTC have a first claim. That is why the $70.9 million receivable line deserves more scrutiny than its size suggests: it is the accounting form that encumbered collateral takes, and it converts what looks like a liquid position into a contractual one. A deposit is a claim on the bank for the same reason — it just rarely gets exercised against a volatile asset that can be margin-called. The $97.6 million quarterly net loss is the figure that will trend. Read its composition before treating it as cash burn. It included $25.3 million in inventory and prepayment write-downs plus purchase-commitment provisions, and $9.2 million in property and equipment impairment. None of that is money leaving the building. It is the cost of equipment that no longer sells at the price it was built to sell at. Cash at June 30 stood at $66 million, up from $43.5 million at March 31, though still below the $80.8 million held at the end of 2025. A rising cash balance next to a nine-figure loss is not a contradiction; it is a reminder that impairment and liquidity are different measurements. Read the mix and the picture does not get better — it gets narrower. Bear-market mechanics compound this. When hashprice compresses, miners carrying equipment inventory take the loss twice: once on the machines they cannot sell at margin, once on the machines they are forced to run below the return their depreciation schedule assumed. The $25.3 million in write-downs and purchase-commitment provisions is that double hit made explicit. The provisions are the forward half of it — contracts signed at a price the market no longer clears. Then came the post-quarter moves. Canaan sold 3,952 ETH and 54 BTC in late August for roughly $13.9 million and directed part of the proceeds into share repurchases. By Sept. 8, buybacks under its existing program totaled about 16.4 million American depositary shares for $7.4 million, including $5.4 million spent in late August. The June treasury snapshot predates all of it. Here is where structural skepticism earns its keep. A miner selling crypto to buy its own stock is not a signal of confidence; it is a capital-allocation loop, and loops have preconditions. The repurchase only makes sense while the discount to book exceeds the cost of liquidating the assets that fund it. Pledged coins cannot be liquidated without unwinding the loan. Fixed-term coins cannot be touched until maturity. Only the 698.5 BTC in the crypto assets category, plus whatever cash the operations throw off, is actually spendable. Model Canaan's treasury as 1,915.5 BTC of dry powder and you have modeled a number, not a balance sheet. The bull framing writes itself: a company trading below the value of its Bitcoin, buying back shares at a discount, mining at a positive cash contribution. The bear framing is the same sentence with one substitution — the Bitcoin is encumbered. Which brings the honest question forward. When a manufacturer's quarterly revenue falls from $42.9 million to $13.6 million and guides lower again, a treasury stops being a strategic reserve and becomes a funding source. Canaan already demonstrated it will convert crypto to cash to retire equity. The next quarter will show how much of the remaining unencumbered stack it is willing to spend on its own shares rather than on the business that generates the shares' value. One disclosure worth watching: paused Ethiopia mining represented nearly 35% of July's operating hashrate. If that capacity stays idle while equipment demand stays soft, the mining cash contribution that made this quarter legible gets harder to reproduce. Q3 guidance is $11 million to $15 million. Watch the receivable line, not the Bitcoin headline. If pledged BTC grows as a share of the total, the treasury is being used as collateral for survival — and that is a different company than the one the multiple implies, and the receivables line will say so first.