Cango Posts $285M Q4 Loss as Bitcoin Mining Costs Surge

Cango Inc Q4 2025 financial results landed Monday with a number that's hard to spin: a $285 million net loss in a single quarter, as impairment charges and fair-value swings on Bitcoin-collateralized positions buried what was otherwise a growing mining revenue line. The company mined more Bitcoin. Spent far more money doing it. And its stock has since cratered over 84% — from roughly $4.50 a share last October to around $0.68 today.

The Q4 Numbers — Revenue Up, Losses Worse

Quarter four revenue came in at $179.5 million, and the lion's share — $172.4 million — came directly from Cango Inc Q4 2025 financial results. On paper, that's a functioning mining operation generating real cash. The problem is what sat on the other side of the ledger.

Total operating costs and expenses for the quarter reached $456.0 million — more than double the revenue. Two items did most of the damage: an $81.4 million impairment charge on mining machines and a $171.4 million loss tied to changes in the fair value of Bitcoin-collateralized receivables. That second figure is the tricky one. When Bitcoin's price moves against you and you're holding collateralized positions, mark-to-market accounting can turn a decent operating quarter into a bloodbath on paper — and that's exactly what happened here.

Chief financial officer Michael Zhang, in a statement accompanying the results, attributed the loss largely to non-recurring transformation costs and market-driven fair-value adjustments. Call it what you want — the math is the math.

The loss was driven largely by non-recurring transformation costs and market-driven fair-value adjustments.

What Does $106,251 Per BTC Actually Mean?

This is the figure that deserves the most attention. Cango's all-in Bitcoin mining cost per BTC rose to $106,251 in Q4 2025 — a threshold that makes profitability essentially dependent on Bitcoin trading well above that level at all times. For context, that's not just electricity and hardware depreciation; it's an all-in figure that absorbs overhead, impairment, and operational scaling costs.

The company mined 6,594.6 Bitcoin across the full year, averaging roughly 18.07 BTC per day. That production rate isn't trivial — but it came at a price. Full-year operating costs hit $1.1 billion, including $338.3 million in impairment losses on mining machines and $96.5 million in fair-value losses on Bitcoin-collateralized receivables.

Scaling mining operations fast — as Cango did after absorbing 32 exahashes per second of capacity from Bitmain in April 2025 — means buying equipment at peak prices, booking depreciation aggressively, and accepting that impairment risk is baked in. The full-year net loss of $452.8 million on $688.1 million in revenue tells that story plainly.

The Bitmain Deal and the Auto-Finance Exit

Cango wasn't always a Bitcoin miner. The company ran China auto financing operations until April 2025, when it agreed to sell that legacy business for $352 million to Ursalpha Digital Limited — an entity linked to Bitmain. Included in that deal: the transfer of the 32 EH/s of mining capacity that turned Cango into a serious mining player almost overnight.

The transformation was intentional and fast. But fast transformations carry transformation costs, and those costs don't always stay neatly in the "non-recurring" column investors are told to ignore.

Is Cango Pivoting Away From Mining?

Here's where the story gets more interesting. In February 2026, Cango sold 4,451 Bitcoin for roughly $305 million, then followed that up with Cango equity financing 2026 totaling $75.5 million — the combination aimed squarely at reducing leverage and shoring up the balance sheet.

The stated reason for that capital raise goes beyond just cleaning up debt. Cango said it plans to repurpose its mining infrastructure into distributed compute capacity for artificial intelligence workloads. Mining rigs running AI inference isn't a new idea — several miners have floated this strategy as Bitcoin mining margins got squeezed post-halving — but Cango appears to be making concrete moves in that direction rather than just mentioning it in earnings calls.

Whether that pivot actually happens at scale, and whether it delivers better margins than mining, is a genuinely open question. AI infrastructure demand is real. But pivoting from Bitcoin ASIC farms to GPU-heavy AI compute clusters is not a simple retooling job. The gap between the announcement and the execution is where many miners have stumbled before.