AI's Power Takeover: How Data Center Demand is Crushing Bitcoin Mining Economics

The core conflict is a direct price war for electrons. AI workloads now command about $25 per kilowatt-hour in value, while Bitcoin mining generates roughly $1 per kilowatt-hour. This massive differential is forcing utilities to prioritize AI data centers, which sign firm, non-stop contracts, over flexible Bitcoin miners. The result is a squeeze on mining economics.

This shift is driving up costs across the board. The U.S. Energy Information Administration forecasts wholesale power prices to rise 8.5% to $51/megawatt-hour in 2026. For miners, whose electricity costs already make up over 80% of operations, this is a direct hit to profitability. As margins shrink, some operators are forced to sell Bitcoin to cover expenses, adding selling pressure to the market.

The impact is visible in price action. Bitcoin has struggled to maintain recent highs, a weakness that aligns with this rising cost pressure. The trend is already visible on the ground, with large-load power requests in Texas surging to 226 gigawatts in 2025, most from AI. This energy reallocation is a fundamental shift, where the cheapest power is being claimed by AI, leaving miners to compete for a smaller, more expensive slice.

Mining's Breaking Point

The financial pressure on Bitcoin miners has reached a critical threshold. The core metric, revenue per terahash, has cratered below 3 cents, making operations unprofitable for all but the most efficient players. This collapse follows a roughly 30% drop over three months to around $28 per terahash per day, a level that would make December's record lows look "enviable."

This economic squeeze is reflected in key profitability indicators. The mining profitability index has hit a 14-month low of 21, while daily mining revenues have fallen to yearly lows of $28 million. The situation is a stark contradiction: post-halving, the industry-wide all-in cost to produce a Bitcoin is estimated between $75,000 and $87,000, while the asset is trading at roughly $64,143. This means the market price is below the average production cost, a condition historically signaling a generational bottom but also a severe strain on cash flow.

The result is a forced industry pivot. Major miners are abandoning pure Bitcoin operations for more profitable alternatives. Companies like Cipher Mining and TeraWulf are shifting to high-performance computing contracts, while others are winding down Bitcoin activities entirely. This strategic exodus is already visible in stock performance, with pure-play miners like Bitmine Immersion Technologies down 29% year-to-date. The financial contagion risk now centers on overleveraged miners and the broader battle for cheap power, as AI data centers outbid them for the remaining low-cost electricity.

The AI Power Play

AI data centers are the undisputed winners in the energy reallocation. The financial incentive is stark: AI workloads generate about $25 per kilowatt-hour in value, while Bitcoin mining earns roughly $1 per kWh. This 25x differential forces utilities to prioritize AI customers with firm, non-stop contracts over flexible miners, directly squeezing mining economics.

This creates a damaging feedback loop. As miners lose access to the cheapest power, their operating costs rise, further compressing already thin margins. This reduces their ability to compete for remaining low-cost electricity, accelerating the industry's pivot to alternatives like high-performance computing. The result is a structural shift where the market's most profitable demand is systematically claiming the grid's best resources.

The scale of this shift is now visible in major infrastructure deals. Nscale's recent acquisition of the Monarch Compute Campus in West Virginia is a prime example. The company plans to build one of the world's largest AI Factories, with a state-certified microgrid capable of scaling to over 8 gigawatts. This project, with initial capacity online by 2028, is being built on a scale that will reshape regional energy markets, securing a massive, long-term power runway for AI compute.