In the past six months, the focus of discussions in the Bitcoin mining community has completely shifted. Previously, everyone was concerned about the price fluctuations of the coin; now, daily questions revolve around how long the support can last. With the coin price hovering around $68,000, mining costs are stubbornly stuck above $80,000, making the math simply impossible to reconcile.

An interesting phenomenon has emerged: miners are not leaving en masse; instead, they are collectively pivoting into the AI computing power sector. The logic behind this shift is worth a close look.

First, let's examine the most striking data. According to the latest industry statistics, the average total cost to produce one Bitcoin across the network has risen to approximately $87,000, while the current coin price is about $68,290. This means that for each block mined, miners are facing an on-paper loss of nearly $20,000. This is not just a thinning profit margin; it’s a complete inversion—producing a coin results in a loss.

Rosenthal Securities’ analysis is even more straightforward: mining revenue has fallen to less than $0.03 per terahash, with only the most efficient mining machines barely breaking even. All others are losing money. The core indicator measuring miners’ profitability—the hash rate price—has dropped about 30% over the past three months, now standing at only $28/PH/day, approaching historic lows.

The result is a wave of large-scale shutdowns. On February 9, the Bitcoin network experienced an 11.16% difficulty adjustment, the largest since 2021, nearly four and a half years ago. Difficulty adjustment is an automatic system protection mechanism that makes it easier for remaining miners to find blocks. But with a 45% cost-price inversion, this adjustment has limited effect.

Within the industry, there is a "Miner Profitability Sustainability Index," with 100 being healthy. Currently, this number is only 21. In plain language: aside from a few top players with electricity costs below $0.05 per kWh and using the latest equipment, most are operating at negative cash flow.

The real shockwave comes from the actions of industry leaders. In late February, Wu Jihan’s Nasdaq-listed company Bitdeer did something unprecedented: it completely liquidated its Bitcoin holdings. Not only did it sell all 189.8 BTC mined that week, but it also sold the 943.1 BTC in its reserves, cashing out roughly $63 million.

This company, which recently became the world’s largest publicly traded self-mining firm with a hash rate of 63.2 EH/s, chose to sell every single coin. In the past, this would have been unthinkable—the belief that "mining equals holding" was a core tenet of the mining community. But now, that belief has been broken.

Bitdeer’s reasoning is pragmatic: with hash rate prices falling below $30, holding Bitcoin entails enormous opportunity costs. Every coin held means less cash available for debt repayment, equipment upgrades, or strategic shifts. They issued $325 million in convertible bonds, explicitly stating the funds would be used for debt buybacks, risk hedging, and investing in AI infrastructure.

Cango has made a similar move, selling 4,451 BTC in early February for $305 million to pay down loans and fund AI computing projects. Former "industry-leading" Bitfarms announced a complete exit from Bitcoin mining, shifting its focus entirely to AI.

On the other side, the story is entirely different. While miners are fleeing, AI companies are pouring in cash.

Morgan Stanley recently calculated that from 2025 to 2028, U.S. data centers’ electricity demand will increase by 74 gigawatts. But what about supply? The new infrastructure under construction adds only 10 GW, and the grid’s available capacity is roughly 15 GW, totaling about 25 GW—leaving a shortfall of 49 GW. This is why mining farms are suddenly in high demand.

What do miners have? Power lines, land, grid connection permits. AI companies’ biggest need isn’t chips—it’s "electricity," specifically "the speed of getting power online." How quickly you can connect power and get your data center running determines how fast you can grab a share of the computing power market.

Morgan Stanley concludes that even if all the available power in U.S. and European Bitcoin mining farms were converted into data centers, there would still be a power shortage. However, retrofitting existing mining farms could reduce the gap by 10 to 15 GW, representing the most efficient "quick fix" solution.

The story flows naturally: mining farms are essentially large-scale, low-latency data centers with power capacity, cooling systems, and rack space. During Bitcoin bear markets, these are cost burdens, but in an era of AI computing power shortages, they become scarce assets capable of generating rental income.

The transition is accelerating. On February 26, one of the largest U.S. miners, MARA Holdings, announced a partnership with Starwood Capital to convert some of its Bitcoin mining facilities into AI data centers. The initial plan is for 1 GW capacity, expandable to 2.5 GW. After the announcement, MARA’s stock surged 17% after hours.

Interestingly, Morgan Stanley’s target price for MARA is only $8, below its closing price at the time. The reason: MARA is not "completely transforming"—it still wants to mine coins and develop AI. The market prefers a "decisive transformation" story.

TeraWulf is a more committed example. The company recently secured financing for a 168 MW AI data center joint venture with FluidStack, with Google as a guarantor of payment. Analysts’ target price for TeraWulf is $37, implying about 159% upside. All 13 analysts have issued buy ratings.

Bitfarms has a more concrete plan: fully converting its 18 MW Bitcoin farm in Washington State to support Nvidia GB300 GPUs, using liquid cooling technology, with completion expected by the end of 2026. They estimate that this site, which accounts for less than 1% of their total capacity, could generate net operating income surpassing all their previous mining profits once converted to GPU-as-a-service.

The valuation logic has completely shifted. Previously, mining company stock prices fluctuated with coin prices and hash rate, like a roller coaster. But if they sign long-term leases and rely on creditworthy payers like Google, cash flow becomes "monthly rent," and the market will value them as infrastructure companies—comparable to REITs like Equinix or Digital Realty, rather than other miners. This approach is called the "REIT Endgame."

Hashrate Index data shows that U.S. miners currently hold about 37.5% of the global market share, Russia 16.4%, and China 11.7%. If U.S. miners shrink their Bitcoin operations to focus on AI, the network’s hash rate will become even more concentrated in Russia and China. For a government that once promised to "make the U.S. the global crypto capital," this is somewhat embarrassing.

But there’s another possible route: selling to allies. For example, Canaan recently spent nearly $40 million to acquire a 49% stake in Cipher Mining’s three Texas farms. These farms have a total capacity of 120 MW, with electricity costs below $0.03 per kWh, and are supplemented with wind power. Capital is flowing, and the computing power landscape is being reshaped.

This wave of "mining to AI" is essentially a swap between two mathematical problems. On the Bitcoin side, halving has cut the block reward to 225 coins per day, with transaction fees unable to compensate; on the AI side, an additional 74 GW of demand is pressing against a 49 GW shortfall. The power, land, and grid permits held by mining farms are shifting from "cost of coin mining" to "hard currency of computing power."

In the short term, shutdowns and conversions will continue. In the long term, whoever can turn "mining volatility" into "rental cash flow" will survive as the next-generation infrastructure company. This industry has never believed in sentiment—only in the "shutdown price."