I just now realized something that significantly changes how we think about Bitcoin miners. Bitdeer has zeroed out its BTC reserves. It went from over 2,400 BTC in November last year and kept selling continuously until it emptied everything around February. Now it only mines and sells immediately.

The curious thing is that the company is showing impressive numbers: revenue of $2.248 billion in Q4 2025, a profit of $705 million, hash rate capacity of 71.0 EH/s with equipment efficiency jumping from 30.4 J/TH to 17.9 J/TH. In other words, operationally, it’s growing strongly. Still, it chose to liquidate everything now. That made me rethink what’s really happening with miners.

What few talk about is that these companies have never truly been Bitcoin accumulators out of conviction. When Bitcoin launched and the mining industry started, the logic was simple: mine, sell, convert to dollars. BTC was a product, not an asset. The accumulation model like MicroStrategy’s became trendy about two years ago, and miners followed the wave, but looking at concrete data, it didn’t stick. A Messari analyst showed that the top 10 public miners between January and November 2025 mined 40,700 BTC and sold 40,300 — a 99% sell rate. No one really accumulated.

The reality is that a mining company is an energy arbitrageur. Bitcoin is just the means to convert cheap electricity into revenue. When the price rises, whether to sell or not becomes a matter of stance. When it falls below the cost of mining, selling ceases to be a choice and becomes a survival instinct.

What’s squeezing the sector now comes from three simultaneous sides. First: the block reward reduction in 2024 cut the reward in half, but electricity costs, depreciation of machines, and operations didn’t decrease at all. Many miners have break-even points above the current price, meaning turning on the machine guarantees a loss. Second: financial numbers are contradictory. Marathon’s revenue grew from $6.56 billion to $9.07 billion but posted a loss of $13.1 billion (had a profit of $5.41 billion the previous year). Hut 8 saw revenue rise from $1.62 billion to $2.35 billion but turned a loss of $2.48 billion (had a profit of $3.31 billion). TeraWulf increased revenue from $1.4 billion to $1.69 billion, but the loss per share in Q4 skyrocketed from $0.21 to $1.66. Revenue is growing but profit is falling across several leaders simultaneously — this isn’t management’s problem; it’s a structural sector squeeze. Third: the macroeconomic environment has become heavy. Trump increasing tariffs, geopolitical uncertainty, risk assets under pressure, BTC dropped significantly. The QCP institution pointed out that Bitcoin is well below the average mining cost. De-risking has become not a strategic option but a real obligation.

Faced with this triple pressure, the only way out miners see is transformation. Using the cheap energy infrastructure and land they accumulated to pivot into high-performance computing and AI. It makes sense on paper: cheap energy contracts and scalable data centers are exactly what’s missing in AI infrastructure. Converting low-margin mining capacity into high-margin computing rental seems like a great business.

Bitdeer is pushing its Sealminer miners, AI cloud services, and HPC businesses. Cipher rebranded from Mining to Digital, announcing a platform transformation. Several are locking in long-term energy contracts at low prices. But real progress is much slower than the narrative suggests. TeraWulf’s HPC revenue was only $970 million in Q4, less than 30% of total revenue of $3.58 billion, and even declined compared to the previous quarter. Acquiring clients, executing contracts, expanding capacity takes time. Debt and share dilution are immediate.

The outcome of this bet depends on whether the new businesses can grow before the debt matures. An interesting point: while BTC fell about 17% in a month, the stocks of several miners rose. TeraWulf went up 31%, Cipher 8%, Hut 8 6%, Core Scientific remained stable. This shows that the capital market is re-evaluating these companies not as leverage for Bitcoin’s price but as potential AI infrastructure providers. The criterion shifted from who holds how many BTC to who locked in cheap energy longer, who has data centers with AI transformation potential, and who has a balance sheet capable of surviving this transition phase.

The truth is that miners have never been the most devout believers in Bitcoin. They are the most rational participants. When mining is profitable, they mine. When accumulating sustains valuation, they accumulate. When selling provides resources for transformation, they sell without hesitation. It’s basic business. The real question now is: when the AI/HPC story is fully priced in by the markets, what will these companies have to show in the next valuation round? If at that moment Bitcoin has already risen but the transformation businesses are not yet mature, will the miners currently divesting assets go back to the story of accumulating coins? Cycles repeat, narratives renew. But in each winter, survival always matters more than faith.