Cryptocurrency mines turn into artificial intelligence factories as twenty-year contracts reshape compute and energy capacity
Listen to this article
Read by Anchor
Global digital infrastructure is undergoing a major shift in power allocation and data centre capacity, as major cryptocurrency mining companies that once filled warehouses with thousands of high-performance machines to mine bitcoin begin repurposing this compute capacity for artificial intelligence workloads and signing long-term contracts with generative model labs.
This shift is driven by declining mining yields and a drop in bitcoin's value from its peak of nearly 124,000 dollars in October 2025. Although it later recovered to around 80,000 dollars, the cost of reversing engineering retrofits makes this transition a one-way path that is difficult to undo once deployed.
Major AI developers, led by Anthropic, are pouring billions of dollars into securing the infrastructure needed to maintain their technical pace. Riot Platforms signed a twenty-year compute agreement with Anthropic valued at 9 billion dollars, while Bitdeer, one of the world's largest mining firms, announced a 16-year deal to supply compute capacity to Anthropic, underscoring the scale of the financial and operational commitments now reshaping cloud hosting.
Operational expertise in running massive data centres and direct access to low-cost electricity have turned mining firms into key strategic partners for AI model developers.
Rebranding has extended to corporate names and websites to reflect these new priorities: Applied Blockchain became Applied Digital, and TeraWulf repositioned itself from a mining operator to a firm focused on high-performance computing and next-generation AI. Enegix, which launched a massive mining facility in Kazakhstan in 2020, has also begun realigning its power capacity and infrastructure toward AI projects, confirming active discussions with hyperscalers to redirect a significant portion of its operations.
Yet retrofitting these facilities is neither simple nor cheap, forcing some firms to sell portions of their cryptocurrency reserves to fund upgrades and prepare server halls for graphics processing units. Energy and computing analysts note that signing ten- to twenty-year hosting contracts provides predictable revenue streams, making decoupling from AI practically impossible even as cryptocurrency prices fluctuate.
For teams building AI systems or managing cloud infrastructure across the Gulf, Egypt and the Levant, this dynamic imposes critical operational realities. Global model labs locking in multi-gigawatt power commitments under two-decade contracts intensifies competition for data centre footprint and electrical equipment. This ensures that leasing advanced compute capacity and GPUs will remain costly, making local power procurement and sovereign hosting facilities essential for regional initiatives seeking to train or run inference without being constrained by global capacity shortages.
In contrast, some operators are turning to a hybrid model, in which cryptocurrency mining offers interruptible loads that can balance grid fluctuations, while AI workloads provide long-term contracts and guaranteed operational revenue that will reshape the data centre landscape for decades to come.