This is another concrete move along the path of mining companies transitioning to AI infrastructure. Over the past year, a batch of companies such as Core Scientific, TeraWulf, and Applied Digital—which were primarily engaged in Bitcoin mining—have begun converting their existing power connections, land, and data center facilities into compute infrastructure for AI training and inference.

Their advantage lies in the fact that many mining operations are already located in regions with low electricity prices and abundant power supply, and have already secured valuable grid connection capacity—which is precisely the most challenging and time-consuming aspect of AI data centers today. Eastern Kentucky's electricity prices and site conditions exemplify this logic of monetizing existing power infrastructure.

For you, these players are both potential competitors and observational samples. The critical question is whether TeraWulf can successfully convert mining assets into high-density, liquid-cooable data centers that meet AI workload requirements, while replicating neocloud's long-term capacity leasing model. The conversion's capital expenditure, PUE performance, and customer contract progress are core metrics for determining whether this path is viable.