Written by: Cathy, Vernacular Blockchain
Mining one Bitcoin costs $87,000. When sold, the market only gives you $67,000.
Each one mined incurs a net loss of $20,000. It's not a loss from transaction fees or fluctuations in electricity costs; it's a solid $20,000 loss every time a Bitcoin is produced. This is the reality as of March 2026. Data from Glassnode and MacroMicro point to the same conclusion: Bitcoin mining, at the current price, is a money-losing business.
However, miners didn't just sit back and wait for their demise. They made a choice that the entire market didn't anticipate—stop mining and sell electricity to AI.
To be precise, it's not that they "stopped mining," but rather they emptied their Bitcoin reserves and invested all their funds into AI data centers, relegating mining to a side business.
Since Bitcoin's decline from its historical high of $126,000 in October 2025, publicly listed mining companies have sold over 15,000 Bitcoins in total. This is not a sporadic cash-out; it's a well-organized, strategic retreat.
Miners collectively liquidate, where did the 15,000 BTC go?
Core Scientific was the quickest and most decisive to act.
In January 2026, it sold approximately 1,900 Bitcoins in one go, cashing out $175 million. The remaining plans aim to clear everything in Q1. This previously bankrupt mining company is now transforming its Texas facilities into high-density AI hosting facilities, targeting to supply all 1.3 GW of total power capacity to AI.
MARA went further. This company, known for "never selling coins," subtly changed its reserve policy in the 10-K annual report of March 2026—53,822 Bitcoins, all authorized for sale. At the then-current price, that's nearly $4 billion worth of chips, turning overnight from "strategic reserves" to "available funds." Shortly thereafter, MARA signed a joint venture agreement with Starwood Capital to provide 1 GW of AI data center capacity.
The most surprising was Cango. Originally a Chinese auto finance platform, it only ventured into Bitcoin mining at the end of 2024 but sold 4,451 Bitcoins in February 2026—60% of its reserves, cashing out $305 million for debt repayment and AI transformation. It also brought in former Zoom executive Jack Jin as CTO of AI business, planning to install containerized GPU computing nodes in global mining sites. A car loan company morphed into a miner in two years, then from miner to AI inference service provider—this kind of cross-domain speed is only seen in the crypto space.
Bitdeer’s choice appears to be a meticulously calculated move. It emptied its Bitcoin holdings in February, and founder Wu Jihan was very frank: holding zero does not mean it will always be zero; liquidity is needed now to seize acquisition opportunities for electricity and land. Unlike other mining companies, Bitdeer accelerated its operations—January Bitcoin production surged 430% year-on-year, and self-owned computing power reached 63.2 EH/s, surpassing MARA, becoming the largest publicly listed mining company in terms of self-owned computing power. The decision to clear out its Bitcoins resulted in significant expansion of computing power and infrastructure. There’s decisiveness akin to “cutting off one’s wrist,” alongside the ambition of “loading ammunition.”
The same electricity, worth 10 times for AI
Why are miners so uniformly selling off? Because after doing the math, the answer is too obvious.
Mining is losing money, but mining companies possess something the whole world is grabbing: electrified land.
After the halving in 2024, the profit margin of Bitcoin mining was compressed from over 90% during peak times to the breakeven point. But during the same period, AI's demand for power and data centers almost skyrocketed. According to MarketsandMarkets, the global AI inference market is expected to grow from about $106 billion in 2025 to nearly $255 billion by 2030.
Morgan Stanley calculated: shifting 1 megawatt of power from mining to AI hosting could achieve a valuation premium of over 10 times.
This is not an exaggeration. AI hosting contracts typically span 10 to 15 years, with clients like Microsoft and Meta, ensuring stable and predictable cash flow. In contrast, mining income entirely depends on coin prices— and you know the volatility of coin prices.
Wall Street has already cast its vote with real money. Morgan Stanley provided Core Scientific with a $500 million loan facility, with the possibility of increasing it to $1 billion. This isn't a loan to a "crypto company"; it's a credit endorsement for a "digital infrastructure company." TeraWulf and Cipher Mining were rated "overweight" by Morgan Stanley due to their successful hybrid models, while MARA, which once clung to Bitcoin, was downgraded due to excessive exposure to coin price risks.
The signals from the capital market couldn't be clearer: in the eyes of Wall Street, these companies' value no longer depends on how many Bitcoins they hold, but rather on how much electricity they control.
On-chain indicators suggest it may soon hit the bottom
Miners are collectively selling off, and the market is in mourning. But if you look at on-chain data, you will find a very interesting set of signals.
The Hash Ribbon inverted at the end of November 2025 and, by February 2026, had persisted for three full months—this is one of the longest miner capitulation periods in history. The last time a similar signal combination appeared was in December 2022 when Bitcoin bottomed at $15,500. As of early March, the 30-day moving average was closing in on the 60-day moving average, with a recovery signal about to trigger.
The MVRV Z-Score maintained between 0.43 and 0.49 in early March. This metric measures the degree of deviation of market price from "real value." Historically, when the Z-Score enters the range of 0 to 1, it almost always corresponds to a strategic accumulation window.
The Puell Multiple dropped to around 0.6, indicating that miners' daily income has been compressed to about 60% of the annual average. It is not far from the 0.3 level of the 2022 bear market bottom, and miners' profit margins are being squeezed to historic lows.
The most extreme signals come from sentiment. During February’s "Bitcoin polar vortex," the crypto fear and greed index plummeted to 5, and on February 5, after a daily adjustment, recorded losses amounting to a historic $3.2 billion.
Four independent indicators lit up simultaneously, and the last time this occurred, Bitcoin was forming a bottom.
Miners selling coins, could it actually be a good thing?
This is the most counterintuitive part of the whole story.
In the past, miner sell-offs were always regarded as a bearish signal—these people are Bitcoin's "native sellers," mining and selling, which constituted a continuous selling pressure on the market. But the nature of the 2026 sell-off is entirely different: these mining companies are selling their Bitcoins to switch to earning income in AI dollars.
Think about what this means. Previously, Core Scientific had to sell hundreds of Bitcoins every month to cover electricity and operational costs. Now, with a long-term contract signed with Microsoft and a credit line from Morgan Stanley, although it still plans to cash out the remaining majority of its Bitcoin holdings (approximately 2,537 by year-end, having sold half), this is no longer a passive "selling to survive." Instead, it’s an active liquidation, concentrating funds into AI infrastructure. Once the MARA and Starwood joint venture project is realized, the dollar cash flow generated by that 1 GW data center will be enough to cover all costs.
In other words, the mining companies transitioning to AI have shifted from structural sellers of Bitcoin to neutral or even potential buyers. The largest group of "natural shorts" in the market is permanently withdrawing.
And Bitcoin mining itself is not disappearing; it has merely transformed into a different existence. MARA's hybrid model has already pointed the way: mine when electricity prices are low, and switch to GPU computing during high demand from AI. Bitcoin has become the grid's "flexible load" and "insurance mechanism," with AI responsible for generating profit and mining ensuring the bottom line.
Conclusion
In 2025, Bitcoin network hash rate had just surpassed the milestone of 1 Zetahash. In the short term, some mining sites transitioning to AI will indeed lead to slower hash rate growth—such as Cango, which has taken 31% of its hash rate offline for upgrades. But this is actually a form of healthy capacity clearance: inefficient miners are exiting, and the remaining players become more efficient and focused, enhancing network security.
This isn't the miners' surrender; it's the evolution of mining.
As mining becomes a side business and AI the main business, Bitcoin loses a group of miners forced to sell coins and gains a healthier supply structure.
Miners may have sold out of their Bitcoins, but the electricity remains.
免责声明:本文章仅代表作者个人观点,不代表本平台的立场和观点。本文章仅供信息分享,不构成对任何人的任何投资建议。用户与作者之间的任何争议,与本平台无关。如网页中刊载的文章或图片涉及侵权,请提供相关的权利证明和身份证明发送邮件到support@aicoin.com,本平台相关工作人员将会进行核查。