In crypto in 2025, survival mattered more than anything else.
While Bitcoin was still whipsawing on the charts and the entire industry was on edge, Canaan released its 2025 report card: annual revenue of $529.7 million, up 96.7% year on year — nearly double.
A year earlier, the company was mired in a gross loss of $84.3 million. Now, its full-year gross profit has swung firmly positive to $41.2 million.
Its full-year net loss still reached $210 million, but that was 15.6% narrower than the $250 million net loss in 2024.
What exactly did Canaan do in just one year?
The Turnaround Was Not Luck
In 2025, Canaan decisively moved away from its old model of living at the mercy of retail buyers and shifted toward institutional customers, pulling off a performance turnaround.
That year, Canaan sold 36.5 EH/s of computing power, with fourth-quarter sales alone surging to 14.6 EH/s, a record high.
The surge in data came from a wholesale change in its customer mix: the company had once depended heavily on retail buyers, but it is now expanding institutional accounts while continuing to push in the retail market.
In July 2025, it signed a fourth purchase agreement with listed mining company Cipher Mining;
in October, it secured its largest single order in three years, covering more than 50,000 units;
and in November, it completed $72 million in institutional financing. The October order, in particular, became the key to reversing its trajectory.
That order pushed Canaan’s product revenue in the fourth quarter of 2025 to $165 million, up 124.5%. One large order locked in revenue; on-time delivery locked in trust.
Canaan’s surge was not a matter of luck. It was the inevitable result of a successful shift in its business model.
But for Canaan, simply selling mining rigs was nowhere near enough.
At the same time, it was doing something else: mining Bitcoin itself.
Mining With One Hand, Hoarding Coins With the Other
In 2025, Canaan’s crypto mining revenue reached $113.2 million, up 157%. The unusual part was that after setting a monthly mining record of 109 Bitcoin, Canaan did not rush to sell.
By year-end, it held 1,750 Bitcoin and 3,951 Ethereum, taking its reserves to a record high.
In the past, the model was “mine coins to pay the power bill.” Now it is “selective selling plus strategic accumulation.”
Its current playbook is clear: sell hardware to generate cash flow and keep the company alive, while using self-operated crypto mining to aggressively build coin reserves as a long-term strategic option.
The confidence behind that approach comes from cost.
According to its CFO, Canaan’s average global electricity cost is just $0.042 per kWh.
That means as long as Bitcoin does not fall below $30,000, Canaan’s crypto mining cash flow remains in the black.
Compare that with peers’ shutdown prices and the logic becomes clearer: assuming an electricity price of $0.06 per kWh, Bitmain’s S21+ has a shutdown price of about $50,000, while the S19 XP is around $66,000;
Canaan’s A16, even by conservative estimates, is only around $30,000. When others can no longer bear the cost and have to shut down, Canaan can keep printing cash.
Of course, there is a price to pay.
Including machine depreciation of more than $10 million in a single quarter, Bitcoin needs to stay at $100,000 to $110,000 for Canaan to achieve a true full-book payback. Hoarding coins is not a guaranteed win either: Bitcoin price volatility in the fourth quarter of 2025 directly caused a $44 million non-cash unrealized loss. It did not break the company, but it was still a close call.
For now, crypto mining and coin accumulation are ultimately just ways for Canaan to defend its core business. The real offensive that will determine its profit ceiling is in Texas.
Control Over Texas Power Is the Real Ambition
As North American miners line up to pivot into AI, Canaan is moving in the opposite direction.
In February 2026, it spent $39.75 million to acquire a 49% stake in three Texas mining farms, known as the ABC project.
On the surface, it was expanding mining farms. In reality, it was buying the 120 MW of power capacity and 4.4 EH/s of computing power behind them.
In crypto mining, computing power is the face of the business. Cheap electricity is the real trump card.
The ABC project — Alborz, Bear and Chief Mountain — sits in the ERCOT power market, where electricity costs are below 3 cents per kWh, among the best levels in the industry.
ERCOT manages about 90% of Texas’s power load and is the only fully deregulated electricity market in the United States.
Its demand-response mechanism allows mining farms to shut down equipment during grid peaks and earn grid-balancing payments.
In other words, when wind power is abundant and electricity prices are low, Canaan can run its machines at full tilt and mine aggressively; when the grid is short of power, it can shut down voluntarily and earn sizable grid-balancing subsidies.
It can profit from both mining and subsidies. But Canaan’s energy strategy does not stop there.
When Canaan converted stranded natural gas in Canada directly into computing power in 2025, achieving electricity costs far below the industry norm, its ambition was already hard to miss.
It is neither blindly pivoting into pure AI data centers nor clinging to traditional overseas mining farms. Its long-term play is to become an infrastructure operator: using the same equipment and energy base, it expects to switch flexibly between crypto mining and AI, while selling excess power back to the grid during curtailment periods.
Over time, it aims to evolve into an energy and computing power infrastructure operator with grid-interactive capabilities.
Razor-Thin Margins and a Cliff-Edge Stock Price
But other numbers in the financial report were far less flattering.
Canaan’s gross margin in the fourth quarter of 2025 was only 7.4%, down 3.6 percentage points from the previous quarter.
The CFO cited three reasons: pricing discount pressure from large customers, higher crypto mining costs, and $13.9 million in inventory write-downs.
Margins that thin leave little room for mistakes, and capital markets were not impressed.
In May 2025, Canaan received its first Nasdaq delisting warning and barely bought time through a 1-for-50 reverse stock split.
A second warning arrived in January 2026. The stock was trading at about $0.79, down roughly 63% over the previous 12 months, with a grace period running until July 13.
The fact that revenue doubled while the company kept flirting with delisting shows that the market cares about whether you can make money, not how much top-line revenue you book.
To save itself, Canaan moved on two fronts. On one side, it raised capital to support the stock: in November 2025, it raised $72 million, and a month later approved a $30 million share buyback, of which about $2 million had been executed as of February 10, 2026.
On the other side, it cut off a limb to survive. In June 2025, it announced the end of its AI chip business, which it had been developing for about 10 years.
That AI chip unit, once seen as a model for transformation, generated only $900,000 in revenue in 2024 while consuming 15% of operating expenses. When CEO Zhang Nangeng spoke of “concentrating resources,” the subtext was clear: the story no longer worked, so it was time to stop the bleeding.
After cutting AI, Canaan still had $80.8 million in cash on its books at the end of 2025, enough to cover its near-term needs.
For now, Canaan has abandoned the seductive story of an AI pivot and put its future back on mining rigs. In 2026, the question is not whether it can grow, but whether that growth can translate into profit and market confidence.
From a chip company in Hangzhou to an energy player in Texas, Canaan has spent a decade taking the long way around.
It has cut AI R&D, bet on cheap electricity, and doubled down on crypto mining.
Canaan has given up on cash-burning AI chip design and turned instead to integrating extremely low-cost electricity. With cheap power in hand, it can keep mining Bitcoin itself, build data centers, and provide computing power hosting for companies building large models.
This is a heavy-asset, high-risk move that later entrants cannot easily copy.
If these heavy assets have not turned into stable cash flow by the end of 2026, the grand narrative on earnings calls will remain nothing more than an empty promise.
Comments
00No comments yet. Be the first to weigh in.