Bitcoin mining companies are moving toward AI to diversify their business and stabilize profits.
The Bitcoin mining sector and the artificial intelligence (AI) sector seem destined for an increasingly close relationship. On one hand, mining companies are searching for greater revenue stability; on the other, AI needs large energy infrastructures. This has led several mining companies to explore the AI data center business. But is this a natural synergy or a competition for energy resources?
The instability of mining revenues
The mining sector has always had to deal with revenue volatility. The probabilistic nature of discovering new blocks, difficulty variations, and bitcoin price fluctuations make it difficult to provide reliable estimates to investors (which is precisely why mining pools exist, to receive more regular rewards).
For example, when a publicly traded company needs to present forecasts for the coming year, it must provide an estimate of its earnings. In Bitcoin mining, such forecasts are difficult to calculate.
Mining companies must face uncertainty by seeking new diversification strategies. Investors in a company want to know the expected return on their investment (ROI), requiring reliable revenue projections and predictable cash flows, elements difficult to guarantee in the mining industry.
AI data center hosting would offer miners an opportunity for stable income: by defining long-term contracts with AI companies, miners can obtain constant cash flows, thus diversifying their business model. This approach would reduce the risk associated with price fluctuations, creating a more predictable source of income.
According to what was stated last August by the CEO of the mining pool Luxor, Nick Hansen, computing operations for artificial intelligence could generate revenues between $2 and $3 for every kilowatt-hour consumed (but Luxor does not specify in which geographic area, ed.), a value significantly higher than mining returns, which stand between $0.15 and $0.20 for the same amount of energy used. Based on an interview with Hut 8 CEO Asher Genoot and third-party estimates, VanEck assumes that AI could generate revenues of $1.30 per kWh.
The attraction of AI data centers
Some of the major mining companies, such as Crusoe Energy, Iren, Hut 8, Cypher Mining, Core Scientific, and Bitfarms, are investing in AI data centers.
The choice is motivated by three infrastructural synergies:
- physical space: both mining and AI require large spaces (large warehouses) to host high-performance hardware;
- cooling systems: both sectors generate a high amount of heat. On one hand, managing the heat produced by ASICs and on the other, cooling GPUs require advanced dissipation systems;
- electricity consumption: both mining and AI are energy-intensive sectors, albeit with different operational needs.
Developing an infrastructure dedicated to AI from scratch involves considerable investments and complexity. For this reason, several AI companies are turning their attention to mining companies, which already have the computational infrastructure necessary to support artificial intelligence workloads. As explained in an August 2024 report by Matthew Sigel, head of digital assets research at VanEck:
“The synergy is a winner: AI companies need energy and Bitcoin miners have it. Bitcoin miners are extraordinarily equipped to immediately support AI [and high performance computing (HPC)].”
Sigel added that some mining companies can be repurposed to serve AI in less than a year, compared to the four years needed for developing AI data centers from scratch.
According to analysts at the financial research and consulting firm Bernstein, led by Gautam Chhugani, by the end of 2027, about 20% of Bitcoin miners’ energy capacity could be destined for artificial intelligence.
According to Yahoo Finance, Jason Les, CEO of Riot Platforms, stated that his company has received several requests from “blue-chip” AI companies interested in securing large energy capacity. In many cases, such companies have offered to cover the capital costs necessary for adapting existing facilities.
“It’s a constant and reliable cash flow,” Les said, adding: “A cash flow that, unlike the rest of our business, is not subject to bitcoin volatility.”
“If you’re collaborating with a well-capitalized partner, you can be sure they’ll honor these agreements for a very long time,” Les concluded.
The energy needs of AI
Artificial intelligence systems and the data centers that host them require considerable amounts of reliable energy. According to the US Department of Energy, data centers consume 10 to 50 times more energy than a standard commercial building.
Estimates from the Federal Energy Regulatory Commission indicate that US data centers will require up to 35 GW by 2030. Experts from the Electric Power Research Institute predict that these centers could consume up to 9% of the entire American electricity production by that date.
Goldman Sachs forecasts suggest a 160% increase in data center energy demand by 2030, mainly due to AI. It is estimated that by 2028, AI will represent 19% of the energy needs of these structures.
Meanwhile, the US Energy Information Administration estimates that bitcoin mining represents between 0.6% and 2.3% of annual US electricity consumption.
Furthermore, while bitcoin mining can adapt to temporary energy interruptions, AI data centers require a constant flow of electricity, which could create competition for electrical supply. However, this very difference in operational requirements could transform into an advantage: miners could act as “load balancers” for energy networks, combining mining with hosting AI workloads. During AI demand peaks, they could reduce mining intensity to free up energy, while during periods of low AI demand, they would increase mining activity.
Revenue diversification
To cope with revenue instability, mining companies have implemented various strategies in the past year.
According to a Clear Street report titled ‘BTC mining: key themes for 2025 emerge‘, miners are pursuing yield strategies for their bitcoin reserves and diversifying into AI/HPC computing (Artificial Intelligence and High Performance Computing). The document outlines three themes for 2025:
- generate revenue on bitcoin reserves;
- leverage existing infrastructure for HPC/AI initiatives;
- benefit from a change in US regulatory leadership.
Some mining companies are looking for new ways to earn from the bitcoin they hold, and one possible option is securities lending. If the SEC allowed direct conversion of bitcoins into ETF shares, miners could exchange BTC for ETFs and earn by lending these shares to institutional investors. According to Clear Street, companies like CleanSpark that hold a large amount of bitcoin could generate millions of dollars in interest annually. Other miners, such as Bit Digital, TeraWulf, and Bitfarms, adopt different strategies, including “staking programs” or immediate sale of bitcoins. A similar approach could open new income sources and improve the sector’s operational efficiency.
Furthermore, the document highlights that many miners are adapting their infrastructures to provide computing power for advanced artificial intelligence applications (HPC), thus diversifying their earnings. The report also emphasizes that the Trump administration could favor mining with more open policies on regulation and energy, with potential tax incentives.
Company investments and strategies
The artificial intelligence sector is experiencing a strong expansion phase, characterized by growing media interest and massive capital inflows. The financial holding Softbank Group, for example, has invested $19 billion in the Stargate project, an AI initiative announced in January by President Donald Trump. This abundance of funding could represent an opportunity for mining companies, not so much as an alternative to their core business, but as a complementary source of revenue. An example is Cipher Mining, which has benefited from a $50 million investment from Softbank.
Crusoe sells mining business to NYDIG
Another example is NYDIG‘s acquisition of Crusoe Energy‘s mining operations. Crusoe, which initially used Digital Flare Mitigation (DFM) technology, exploiting the phenomenon of gas flaring from oil fields, has decided to focus on building optimized AI data centers, leaving Bitcoin mining in the hands of NYDIG. About 135 Crusoe employees will join NYDIG and continue to manage the business under the direction of the new company. No roles will be eliminated as a result of the transaction, although specific terms have not been disclosed. NYDIG plans to continue operating and investing in the growth of the business.
Regarding the sale, Crusoe Energy CEO Chase Lochmille said:
“We will continue to take the same energy-focused approach to power AI infrastructure and accelerate its adoption and diffusion in everyday life.”
Over the years, their mining operation has implemented more than 425 modular data centers (over 250 MW) in seven US states and in Argentina. Since the technology’s launch seven years ago, Crusoe’s DFM business has mitigated 2.7 million tons of greenhouse gas emissions and prevented nearly 6.23 billion cubic meters of natural gas from being burned—equivalent to taking nearly 630,000 cars off the road for a year.
Core Scientific: from bankruptcy to comeback
In October 2024, Core Scientific, a bitcoin mining company, announced an expansion of its partnership with CoreWeave, a Nvidia-backed cloud computing provider, to rent space for AI-based services. The agreement, worth $8.7 billion over 12 years, includes providing an additional 120 megawatts of computing infrastructure to support CoreWeave’s operations. In total, CoreWeave has agreed to rent 500 megawatts of space.
The company’s transformation is particularly relevant considering that in January 2024, Core Scientific had filed for bankruptcy under Chapter 11. After emerging from bankruptcy proceedings on January 23 with $400 million less debt, the company started the year focusing entirely on mining, then quickly changed direction after seeing increased demand for electricity for AI-dedicated data centers.
Adam Sullivan, CEO of Core Scientific, emphasized how the company’s existing infrastructure has allowed it to diversify revenues and reallocate some facilities to respond to the growing demand for technology necessary for AI.
Bitfarms diversifies with AI data centers
Last January, the Toronto-based mining company Bitfarms commissioned two consulting firms in HPC/AI, Appleby Strategy Group and World Wide Technology, to explore how it could transform some of its facilities to meet the growing demand for AI computing resources. CEO Ben Gagnon emphasized that HPC/AI contracts guarantee stable cash flows, while Bitcoin mining will continue to offer growth opportunities.
Last March, the company completed the acquisition of Stronghold Digital Mining, securing an expansion process of 1.1 gigawatts distributed across three sites in Pennsylvania. The operation directly supports Bitfarms’ strategy to expand its presence in the USA and develop a large-scale high-performance computing (HPC) and artificial intelligence business.
On April 2, the company announced obtaining private financing of up to $300 million from Macquarie Equipment Capital to develop its HPC data center in Panther Creek, Pennsylvania.
Hut 8: financing for AI development
Last June, the Canadian company Hut 8 secured $150 million from private equity firm Coatue Management to invest in AI. The financing took place through convertible notes with an annual interest rate of 8% and a conversion rate of $16.395 per share.
TeraWulf: rental to AI companies
On December 23, 2024, the mining company TeraWulf announced the rental of more than 70 megawatts of its data center infrastructure to the AI and cloud company Core42. This decision marks an expansion of TeraWulf’s activities, which now also includes hosting service for AI.
Iren: expansion into AI
On March 31, the mining company Iren, previously known as Iris Energy, announced a strategic change in its business operations, deciding to suspend the expansion of Bitcoin mining to 52 EH/s to focus on developing AI and HPC infrastructure.
The company currently operates on three main fronts:
- Bitcoin mining with capacity increased from 31 EH/s to 35 EH/s, which generates a projected annualized net cash flow of $528 million;
- AI Cloud services with 1,896 GPUs in operation, generating annualized revenues of $26 million as of March 31, 2025;
- AI Data Centers, with the development of Horizon 1 (50 MW load) requiring an investment of $300-350 million, and the Sweetwater Data Center Hub project with a power capacity of 2 GW.
Daniel Roberts, co-founder and co-CEO of Iren, stated:
“As we approach the completion of our mining expansion to 50 EH/s, our focus is shifting toward the next phase of growth and providing scalable infrastructure for AI and HPC through our AI Cloud Services and AI Data Center businesses.”
While on one hand mining companies might find new sources of income by exploiting the growing demand for AI infrastructure, on the other hand, competition for energy could intensify.
According to what was stated to Atlas21 microphones by Paolo Ardoino, CEO of Tether:
“The coexistence between bitcoin mining and AI will be difficult. What I think will happen is that the two sectors will compete for access to capital, access to chips, and access to energy, and unfortunately, AI at this moment enjoys greater attention in the eyes of the global community and in the eyes of investors. Despite both technologies being exciting and potentially very compatible, I think the gold rush on AI will make things more difficult for Bitcoin miners.”
