The market says Canaan is worth $144.7 million. Canaan's own balance sheet says the sum of its digital-asset reserve and its cash already reaches $173.5 million. That gap โ roughly $28.8 million, or 19.9% of the Aug. 4 intraday market capitalization โ is the kind of discrepancy that makes value investors reach for their calculators and short sellers sharpen their pencils.
Then came the kicker hidden inside the Aug. 4 SEC-filed announcement. Canaan has opened a new funding channel for its share-buyback program: management can now use proceeds from its roughly $130 million digital-asset treasury to repurchase its own stock. Not a fresh authorization. Not a new vehicle. The same 12-month buyback program that began on Dec. 12, 2025, is now rear-loaded with a brand-new funding source โ the company's own Bitcoin and Ethereum. Execution details remain undisclosed on both sides of the trade, from any treasury sale to any subsequent repurchase.
Let me translate that into human terms, because the legalese obscures the gravity. Canaan, a Nasdaq-listed Bitcoin-mining hardware maker that lost $88.7 million last quarter, has just authorized itself to sell the coins it hoards in order to buy the shares it deems undervalued. The vault is being opened from the inside.
Let's be clear about the stakes. This is not a software company burning excess cash. This is a hardware maker whose only differentiated asset in a brutal ASIC market is the balance sheet itself. The moment the balance sheet becomes the fuel for share-price defense is the moment investors should stop treating the treasury as a valuation anchor and start treating it as a consumable.
Speed is the currency, but accuracy is the vault. So let me be precise about who Canaan is and why this filing matters a lot more than a routine buyback footnote.
Canaan makes Avalon ASIC miners. It went public through a convoluted backdoor listing, and it has spent the past two years riding a hardware cycle that has been brutal to its specific corner of the Bitcoin economy. The 2021 mania created a global appetite for machines; that appetite is now a memory. First-quarter results showed a gross loss of $22.9 million, an operating loss of $54.3 million, and a net loss of $88.7 million. The company's own $35 million to $45 million second-quarter revenue guidance was described as 'in line' โ translator: the collapse has slowed, but it hasn't stopped. The competitive landscape makes the timing worse. Canaan is fighting Bitmain and MicroBT for a shrinking pool of hardware orders, and ASIC price declines have been unrelenting. When the machine you sell is a commodity with rapidly decaying profitability, the only moat is cost structure โ and Canaan's cost structure is not improving while its cash pile shrinks.
The buyback program itself has been a patient animal. Launched Dec. 12, 2025, it carries a 12-month ceiling of $30 million for ADS or Class A ordinary share repurchases. By May 19, Canaan had spent roughly $2 million to buy back about 2.8 million ADSs. Simple subtraction leaves a nominal capacity of about $28 million as of May 19. The Aug. 4 release did not disclose a newer repurchase total, so the current unused authorization is officially unknown. But the shape of that number is unmistakable. Twenty-eight million dollars of dry powder against a $144.7 million market capitalization is nearly one-fifth of the entire public-market value of the company. And the freshly disclosed funding channel lets management pull that one-fifth directly from a digital-asset treasury that held 1,915 BTC and 3,952 ETH as of June 30, according to the company's June operating update.
Now look at the coincidence that should make any surveillance-minded analyst pause. The 19.9% gap between Canaan's market cap and the gross sum of its treasury-plus-cash roughly equals the size of the remaining buyback authorization. The market is not just discounting the business. The market is pricing the likelihood that management will reach into its own pockets, pull out the value, and prove the market wrong.
Based on my experience tracking balance-sheet footnotes since the 0x-protocol days, I would argue the filing is a signal of something more fragile than a routine capital-return program. It's a confession.
Let me take you into the balance-sheet autopsy, because the flashy $130 million figure is not the same as $130 million of spendable money. The June operating update listed 1,915 BTC and 3,952 ETH on the balance sheet as of June 30, including receivables and excluding customer deposits. The May earnings coverage put that treasury near $148 million at the time. By Aug. 3, the estimate had dropped to approximately $130 million. That is an $18 million mark-to-market loss in roughly ten weeks. The coins have already begun melting before a single buyback trade is printed.
Even the headline number deserves a closer read. The June update's 1,915 BTC and 3,952 ETH include receivables โ coins that customers owe but Canaan has not yet collected โ and exclude customer deposits, which are coins Canaan holds as a liability rather than as equity. That accounting distinction matters in a hardware cycle where disputes over undelivered machines are rising. The 'treasury' is not one homogeneous vault; it is a stack of claims with different degrees of collectability and different legal priorities.
The restricted layers make it worse. As of March 31, Canaan had pledged 905 BTC for secured term loans and placed another 100 BTC into a fixed-term product. Those March figures predate the June holdings, so the latest restriction status is uncertain โ and that uncertainty itself is a red flag. But run the directionally reasonable math at early-August implied prices, with Bitcoin in the low $60,000s and Ether in the high $2,000s: the 905 pledged BTC are worth somewhere north of $55 million, and the 100 locked BTC add roughly $6 million more. That is over 1,000 of 1,915 coins โ 52.5% of the entire BTC stack โ that was already spoken for, pledged against debt, or time-locked before this new filing arrived.
Even the unencumbered leftovers don't breathe easy. The March 31 balance sheet showed $43.5 million of cash against $106.4 million of current liabilities. Cash was $80.8 million at the end of last year; the liability wall has stayed high while the cash pile shrank by $37 million in a single quarter. The roughly $42 million of customer-receivable collections in April were a lifeline, sure, but collections are the echo of past sales, not a signal of future demand. In my experience dissecting miner balance sheets across the 2017-to-2022 cycles, receivables in a bear market are the boomerang of an earlier boom โ they pay for your past, never for your future.
Run the simplest liquidity ratio and the picture darkens further. $43.5 million of cash against $106.4 million of current liabilities gives a current ratio of roughly 0.41. A company with a current ratio below 1 is already depending on its inventory, its receivables, and its secured lending relationships to make it through the quarter. Now contemplate what a $28 million buyback funded by the digital-asset reserve does to that ratio: it converts one of the most liquid assets on the book into a retired share count. Working capital decreases. The company survives only if the hardware cycle turns before the cash runs out โ or if the board keeps treating the coins as a piggy bank.
That is why the headline discount is a directional illusion. Combining an Aug. 3 digital-asset estimate with a March 31 cash balance creates a $173.5 million gross sum, but it omits liabilities, asset restrictions, and the minor detail that 52.5% of the Bitcoin isn't free to sell. Same-date net asset value would require a fuller balance sheet. Anyone who tells you Canaan is trading at a clean 20% discount to asset value is telling you only the part of the story that flatters the trade.
The more important question is what a crypto-funded buyback actually does to net asset value per share once it executes. This is the piece nobody on the tape is talking about. If management sells roughly $28 million of treasury coins to buy back roughly $28 million of stock at prices close to current marks, it removes nearly equal value from both sides of the equation. The per-share NAV barely improves, because the denominator shrinks at the same rate as the numerator. The buyback isn't closing the discount; it's rotating it. Coin value becomes share-buyback value, and the discount on the balance sheet becomes a discount on the share count. That is not value creation. That is asset conversion with extra steps.
Here is the contrarian angle nobody has framed yet: using crypto proceeds for buybacks is not a shareholder-friendly unlock. It is a capital-allocation confession. When a company that manufactures Bitcoin miners decides its own shares are a better buy than Bitcoin itself, it is telling the market โ from a seat with a direct view of the order pipeline โ which asset it expects to outperform from here. And it is not the coin.
The pure per-share argument is real: a well-timed repurchase can support per-share value. But each dollar directed to buybacks shrinks the reserve available to fund a loss-making mining and hardware business. Canaan's working-capital needs and board oversight set the limit, and the current working-capital picture is the opposite of reassuring. With $106.4 million of current liabilities and an $88.7 million quarterly net loss, every dollar of treasury that crosses the table to buy an ADS is a dollar that will not be there for the next covenant test, the next inventory order, the next payroll.
And do not mistake the mechanism for a hedge. If Bitcoin rallies, Canaan will have missed the upside on the coins it already sold to fund the buyback. If Bitcoin falls, the 905 pledged BTC become collateral under pressure, and the lenders will come looking before the buyback team does. Trading a volatile asset you own for your own equity is not volatility reduction; it is volatility substitution. You remove one coin from the balance sheet and insert a more junior claim on a loss-making business.
There is also an ugly symmetry for the remaining shareholders. Every ADS retired by a crypto-funded buyback increases the proportional claim of the survivors โ on a business with a shrinking balance sheet. That is not the same as increasing proportional claims on growing earnings. In a normal company, buybacks retire excess capital. In Canaan, the buyback is retiring the capital itself. The remaining shareholders end up owning more of a company that owns less. The per-share math can look accretive while the absolute value of the enterprise quietly dissolves.
Compare this with Strategy, which in May put Bitcoin sales on the table for its own repurchase program. The bigger risk identified there was that the pile is now explicitly listed as a funding source โ the treasure becoming a float. But Strategy is fundamentally a leveraged Bitcoin vehicle; its operating model is treasury management by other means. Canaan is not. It is a hardware vendor with no power contracts, no mining sites, no AI-rerating story, and no way to join the pivot that CoinShares documented this spring, when stronger operators moved into AI while stressed miners sold coins to stay liquid. Canaan cannot join the AI camp because it does not control the infrastructure; it cannot wait out the cycle like a pure miner because its revenue dies the moment customers stop ordering machines. So it reaches for the only lever left: buying its own stock with its last liquid coins.
Echoes of 2017 whisper through every new bull run. I was on-chain during the ICO mania, triangulating order flow across 0x relayers when everyone else was staring at token prices on CoinMarketCap. The 2017 pattern was simple: treasury hoards look like a safety net until the operating business eats the net. The projects that survived were the ones that could wait. The ones that sold their own reserve to defend a token price simply accelerated their irrelevance. Canaan is now running that exact play, but with a Nasdaq ticker and a quarterly net loss instead of a whitepaper.
And nobody is talking about the 905 pledged BTC. Those coins are not in a cold wallet awaiting a redemption campaign; they are collateral. If Bitcoin drops another significant leg, the lenders behind those secured term loans start looking at that collateral with a very different set of assumptions. The treasury-funded buyback quietly becomes treasury-transferred-to-lender. That tail risk is the most under-priced part of this entire setup โ and it cannot be fixed by buying back stock.
The data-availability narrative of 2025 taught a lesson that applies directly here. Ninety-nine percent of rollups do not generate enough data to justify a dedicated DA layer, and the industry built an entire infrastructure narrative anyway. Canaan has the same problem in reverse: it does not generate enough earnings to justify a dedicated buyback war chest funded by its own crypto. The SEC program, the funding channel, the quarterly reporting theater โ it is all wildly out of proportion to the fundamental problem, which is that the machines are not selling and the losses are compounding.
So here is what I am watching, and what you should be watching too. First, the next 10-Q: the buyback run-rate matters more than the headline authorization. A $2 million spend over five months is a trickle. If the crypto channel accelerates that trickle into a real outflow, you will see it in the ADS count within two quarters โ and you will see the BTC line shrink. Second, the pledged-coin status: whether those 905 BTC remain pledged matters more than the price of Bitcoin. Third, the operating cash burn: can the hardware business produce positive working capital on its own, or is it structurally dependent on the treasure melting?
And watch for copycats. If Canaan's play works โ if the share price stabilizes and the market applauds the financial engineering โ other listed miners with decimated valuations and idle treasuries will read the same playbook. The risk is a wave of crypto-funded buybacks that systematically converts the sector's digital reserves into shrinking share counts, hollowing out the very asset base that gives mining equities their reason to exist. The first company to do it is an anomaly. The fifth is a trend. And trends in a bear market are how balance sheets get emptied.
And beneath all three, the real question no single filing can answer: in a bear market, survival matters more than gains. Canaan's board has decided that its stock is the most undervalued asset on its own books. But the books as they actually stand โ $173.5 million gross assets, $106.4 million of current liabilities, $88.7 million of quarterly net losses, 52.5% of the Bitcoin already encumbered โ keep telling a different story. The buyback could be the most disciplined capital decision management makes this year. Or it could be the last deployment of a reserve that was supposed to keep the company alive through the cycle.
The ledger does not care about funding channels or authorized programs. It only records who is buying, who is selling, and in which direction the value flows. Right now, directionally, the flow is from Canaan's coin vault into Canaan's own stock. Management is betting roughly $28 million worth of discount โ between where the market prices the company and where the balance sheet says it should be โ that they know something the market misses. Maybe they do. Maybe the ASIC cycle turns at exactly the right moment, and this becomes a footnote in a turnaround story.
But I have been watching this industry long enough to know: when the coins start flowing out of the treasury to pay for anything other than survival, it is not a signal of strength. It is a last meal. Speed is the currency, but accuracy is the vault โ and in this trade, the vault is being drained from the inside. Watch the next disclosure like your position depends on it. Canaan's does. The tape doesn't lie; the footnotes do.