Transitioning cryptocurrency mining sites into AI data centers is far from straightforward, according to Phil Harvey, CEO of Sabre56. One of the biggest hurdles is cost. While running a commercial crypto mining facility typically costs between $300,000 and $350,000 per megawatt, AI data centers require significantly more capital—between $3 million and $5 million per megawatt. This represents a 10- to 15-fold increase in expenses, making it challenging for mining companies to easily pivot to AI operations.
Moreover, only a small portion of current mining infrastructure is suitable for conversion to AI or high-performance computing (HPC) data centers. Harvey estimates that only about 20% of a typical miner’s portfolio could be repurposed for such uses. This is because AI data centers have far more stringent requirements, including much larger physical space—about 5,000 square feet per megawatt, compared to 1,000 square feet for crypto mining.
In addition to spatial and cost concerns, there’s the issue of replacing the existing infrastructure. Up to 90% of the current equipment used for crypto mining would need to be swapped out, leading to significant upfront investment. As a result, while the idea of transitioning to AI data centers is appealing, especially given the rising demand for AI and HPC services, the process is highly complex and costly for mining companies.
These challenges suggest that while mining firms are exploring new revenue streams, such as AI data centers, the shift is far from seamless and requires substantial financial and operational overhauls.