More and more publicly listed Bitcoin miners are shifting capacity into data centers for artificial intelligence and high performance computing. They bring exactly what the AI market urgently needs: secured access to power, established sites with ready grid connections, and years of experience operating energy-intensive facilities. Why the industry is being revalued and what this means for the market.
Key Takeaways
Numerous US-listed Bitcoin miners are now focusing on AI and HPC. According to Bernstein, 19 AI and HPC contracts covering around 7 gigawatts were signed within two years, with a total value of over 135 billion dollars.
The transferable value lies in the infrastructure: secured power access, ready sites with grid connections, and operational experience.
AI hosting complements price-dependent mining revenues with predictable, dollar-denominated income with terms of up to 20 years.
The stock market rewards the transformation: mining stocks gained over 50 percent by early June, and the market increasingly values these companies like data center operators.
The transformation is capital-intensive and carries risks: high construction costs, returns that often fall short of expectations, rapidly aging GPU hardware, and dependence on major tenants.
For the Bitcoin network, the trend means declining hashrate and difficulty, from which remaining miners can benefit through reduced competition.
A New Chapter for Bitcoin Miners
The Bitcoin mining industry is undergoing a transformation. More and more companies that previously supplied computing power primarily for the Bitcoin network are building capacity for artificial intelligence and high performance computing.
The scale can be quantified: According to a Bernstein analysis, Bitcoin miners have allocated 19 contracts covering around 7 gigawatts of power capacity to AI and cloud companies within two years, total value: over 135 billion dollars. And this is likely just the beginning, as around 30 gigawatts are planned across the sector.
The most prominent deals include TeraWulf's 20-year lease with the AI company Anthropic worth around 19 billion dollars, IREN's five-year contract with Microsoft worth 9.7 billion dollars, and Cipher's 15-year contract with AWS worth 5.5 billion dollars. Core Scientific, the pioneer of the model, holds contracts with CoreWeave worth more than 10 billion dollars, and most recently CleanSpark joined with a 6.6 billion dollar deal.
The stock market rewards the strategic shift. Mining stocks gained over 50 percent by early June, according to 10X Research. CoinShares expects listed miners with AI contracts to generate up to 70 percent of their revenue from the AI business by year-end. The market increasingly values these companies like operators of digital infrastructure.
The Terms in Detail: AI, HPC, and Bitcoin Mining
Let's first take a closer look at the terms: What is behind AI, HPC, and Bitcoin mining, and where do their differences and similarities lie?
AI Infrastructure
Artificial intelligence refers to computer systems that solve tasks normally requiring human thinking, such as writing texts, recognizing images, or understanding speech.
Use cases: Chatbots like ChatGPT, image generators, translation tools, voice assistants, medical diagnostics, autonomous driving, or AI applications in businesses.
Hardware: Graphics processing units (GPUs) that can perform many calculations simultaneously.
Requirements: Very high power density in a confined space, powerful cooling, and highly reliable operation, as these facilities consume significantly more energy per area than conventional data centers.
HPC Infrastructure
High performance computing refers to solving extremely compute-intensive tasks that a normal computer cannot handle. The difference from AI: HPC calculates with predefined rules and formulas, such as physical equations, rather than learning patterns from data.
Use cases: Weather forecasts, climate models, scientific research, drug development, crash test simulations in the automotive industry, or risk calculations in finance.
Hardware: GPUs or specialized accelerators, similar to AI.
Requirements: High computing power, stable energy supply, and resilient operation.
Bitcoin Mining Infrastructure
Bitcoin mining refers to providing computing power to confirm transactions on the Bitcoin network and secure the network. As a reward, miners receive newly created Bitcoin and transaction fees.
Use cases: Exclusively the Bitcoin network, meaning the processing of transactions and securing the blockchain.
Hardware: ASIC miners, specialized chips that master a single computing task (SHA-256 calculations).
Requirements: Low-cost power, fast deployment, and simple cooling. Brief interruptions and remote locations are not a problem, making mining significantly more flexible than AI or HPC.
The Comparison: What Can Be Transferred and What Cannot
The Differences
The decisive point is that mining hardware cannot simply be repurposed for AI. Anyone switching to AI must procure entirely new computing hardware and usually convert the cooling to liquid cooling. AI racks have a much higher power density than mining rigs and require a high-performance, low-latency network connection. Added to this are significantly higher requirements for availability and service quality, as paying AI customers depend on reliable operation. The capital requirement for GPUs is also enormous. A site that has power is therefore far from being a finished AI data center.
The Similarities
All three workloads need more than just chips. They require power, cooling, physical space, network connectivity, and professional, fail-safe operation. Precisely this combination is scarce and cannot be created overnight. The transferable value therefore lies in the surrounding infrastructure, not in the machines: power connection, site, buildings, cooling, and the expertise to operate energy-intensive facilities.
Why Bitcoin Miners Are Well Positioned for AI and HPC
Four factors give mining companies an edge in the race for AI capacity.
1) Access to Power
Mining companies often have large power capacities or long-term energy contracts. In the AI infrastructure market, secured access to electricity is becoming one of the most critical bottlenecks of all. Those who already have this access possess exactly what AI demand is looking for.
2) Strategically Advantageous Locations
Many mining facilities are located in regions with particularly low energy prices, for example near hydropower plants or surplus grid capacity. This cost advantage persists when operating AI and HPC workloads and makes such sites attractive to data center tenants.
3) Operational Experience
Mining companies have been operating energy-intensive facilities for years. They know how to handle hardware, monitor performance, and optimize cooling. Added to this is a mindset of fast deployment, modular infrastructure, and strict cost control, which is also in demand in the AI buildout.
3) Faster Time-to-Market
New greenfield AI data centers often go through years of planning, permitting, and construction. Existing mining sites already have land, buildings, and grid connections, significantly shortening the time to operational infrastructure.
Why Miners Are Diversifying Their Business Models
The trend is not purely an opportunity, but also a response to economic pressure. Three developments explain why the industry is broadening its revenue streams.
The starting point is the earnings situation in the core business: Revenues in Bitcoin mining move with the Bitcoin price, difficulty, and electricity costs. In strong market phases, the business is highly profitable; in weaker ones, margins decline. A second pillar can balance out these fluctuations.
This is exactly where hosting and colocation contracts come in. They deliver fixed, dollar-denominated income with terms of up to 20 years. On the tenant side are creditworthy tech corporations and AI cloud providers, and in some deals their prepayments cover part of the construction costs. Such contracts complement mining revenues with calculable cash flow and make revenue planning more independent of the market cycle.
This stability has not gone unnoticed by the capital market. The market increasingly sees mining companies not as pure crypto miners, but as operators of power and data center infrastructure. This revaluation is reflected in higher stock prices and valuation multiples. It also facilitates access to capital: Those who can demonstrate long-term contracts with financially strong tenants receive debt capital on better terms and can finance further expansion more easily.
The Risks of the Transformation
1) High Capital Requirements
AI infrastructure is many times more expensive than mining infrastructure: Those building themselves invest 8 to 11 million dollars per IT megawatt, compared to 700,000 to one million dollars for Bitcoin mining, according to CoinShares. The buildout is financed through debt and partly through the sale of companies' own Bitcoin holdings.
2) Returns Below Expectations
The headline billions mask the actual returns. According to a Bernstein analysis, the stabilized returns on capital of most colocation deals are in the single digits.
3) Technology Risk
GPU hardware ages quickly, requiring ongoing reinvestment. Unlike mining, where older machines can keep running on cheap power, AI customers expect state-of-the-art hardware.
4) Dependence on Major Tenants
Many operators depend on individual contract partners: If a major tenant defaults or does not renew, large amounts of capacity are left without a buyer. Current valuations already price in that the contracted billions will actually turn into built and profitable data centers.
What This Means for Bitcoin Mining
The trend also feeds back into the Bitcoin network. Those shifting capacity to AI withdraw computing power from the network. For the first time in six years, the hashrate declined in a first quarter, and the share of North American pools in Bitcoin blocks fell from 40 to 35 percent in 2025. In mid-June, mining difficulty dropped by 10 percent, already the second decline of this magnitude in 2026. The security of the network is not endangered by this, but the shift is structural in nature: Sites once converted for AI are unlikely to return to mining even in the next bull market. For the remaining miners, this means less competition: When difficulty falls, the probability of finding a block increases, and the yield per unit of computing power improves. Those who continue mining efficiently and with low-cost power can benefit from the consolidation.
Sources
https://cryptonews.net/news/mining/32977167/
https://www.kucoin.com/blog/Why-BTC-Miners-are-Pivoting-to-AI-Data-Centers-in-2026
Frequently asked questions
Why are Bitcoin miners moving into AI and HPC?
Because the AI boom is generating enormous demand for computing power, and miners possess exactly the infrastructure that is being sought: secured power access, ready sites with grid connections, and experience operating energy-intensive facilities. While new AI data centers go through years of planning and permitting phases, mining sites can be converted significantly faster. In addition, attractive long-term contracts with financially strong corporations are a draw.
Can Bitcoin mining hardware be used for AI?
No. ASIC miners are built exclusively for SHA-256 calculations and are unsuitable for AI workloads. AI and HPC run on GPUs or specialized accelerators. What is transferable is therefore not the hardware, but the surrounding infrastructure: power connection, site, buildings, cooling, and the expertise to operate energy-intensive facilities. Anyone switching to AI must procure entirely new computing hardware and usually convert the cooling to liquid cooling.
Is AI more profitable than Bitcoin mining?
The two models cannot be compared directly because their returns work differently. Mining revenues are market-dependent: In bull markets, high margins are possible; in weak phases, they decline significantly. AI revenues are more consistent because they run on long-term contracts at fixed terms, but without the upside potential of a rising Bitcoin price. Mining thus offers more upside potential with stronger fluctuations, while AI delivers predictable, steady returns.
Will AI displace Bitcoin mining?
Based on current assessments, not entirely. Most companies combine both, using AI hosting for stable cash flow while holding on to mining for its price potential. According to industry reports, individual companies have announced plans to phase out mining in the medium term. Whether this radical course will prevail remains open. More likely is a coexistence in which miners become hybrid data center operators that continue to mine Bitcoin.
Why do AI companies work with Bitcoin miners?
Because miners offer quickly available, power-supplied sites that are otherwise hard to come by in the AI buildout. Instead of planning new data centers for years, AI and cloud providers rent capacity from operators who have already secured power and land.
What is the biggest hurdle in converting a mining site into an AI data center?
Usually the technology beyond the power connection. AI racks have a much higher power density than mining rigs and require advanced cooling, often liquid cooling, as well as a high-performance, low-latency network connection. Added to this are significantly higher requirements for availability and service quality, as paying AI customers depend on reliable operation. The capital requirement for GPUs is also enormous. A site that has power is therefore far from being a finished AI data center.
Are all mining sites suitable for AI and HPC?
No. The decisive factors are grid connection, potential latency, cooling, land, and proximity to fiber. Some mining facilities are deliberately located in remote regions with cheap power but weak connectivity, making them unattractive for latency-sensitive AI workloads. Other sites with strong grid and data connectivity are well suited. The conversion also requires high investments. Only a portion of existing capacity is likely to be sensibly repurposed.
