TeraWulf has bought a Kentucky site designed for more than 1 gigawatt of AI and high-performance computing capacity, a move that makes the company’s pivot away from pure Bitcoin mining impossible to ignore.
Summary
- The Muskie Data Campus in eastern Kentucky is expected to support more than 1 GW of data center capacity on a roughly 285-acre parcel inside a 1,000-acre industrial park.
- TeraWulf says the first 500 MW should begin delivery in the second half of 2028, with a second 500 MW phase targeted for the second half of 2030.
- The company reported $34 million in Q1 revenue, including $21 million from HPC leasing, showing AI compute has already become more important than hashprice-driven mining revenue.
TeraWulf disclosed Tuesday that it has acquired the Muskie Data Campus from Industrial Equity Partners, adding a hyperscale AI and HPC site in eastern Kentucky that the company says can support more than 1 GW of future data center capacity. The property sits on about 285 acres within the broader 1,000-acre EastPark industrial complex, where site work has already started and much of the zoning and permitting framework is reportedly in place.
Power access is the real story. Kentucky Power, an American Electric Power subsidiary, is building a 345 kV substation that will connect into an existing 765 kV transmission network, giving TeraWulf the kind of grid-scale interconnection that matters far more in the AI buildout than old crypto talking points about “hashrate leadership.” According to the company’s filing, the first 500 MW tranche is expected in the second half of 2028 and the second 500 MW tranche in the second half of 2030, making this a long-dated infrastructure bet rather than a quick speculative headline.
The acquisition also expands TeraWulf’s Kentucky footprint to more than 2.8 GW across two projects, strengthening its argument that it is no longer just a Bitcoin miner with side ambitions in compute. That matters because the firm’s economics are already changing: in the first quarter of 2026, TeraWulf generated $34 million in revenue, of which $21 million came from HPC lease revenue, while Bitcoin mining contributed less than $13 million.
This is a compute company wearing a miner’s skin
That revenue mix is the key fact, and it cuts through the marketing. TeraWulf’s AI and HPC leasing business has already overtaken mining as its primary top line, which means the company is effectively monetizing power access, land, cooling and interconnection rather than relying mainly on block rewards and transaction fees.
In a previous crypto.news report, TeraWulf’s first-quarter results were described as the first moment when high-performance compute clearly surpassed Bitcoin as the company’s core revenue driver. Another crypto.news article noted that Bernstein sees more than $90 billion of announced AI infrastructure partnerships and argues that listed miners now control over 27 GW of planned power capacity, turning them into strategic gatekeepers for compute rather than just operators of speculative hash factories.
That said, the market should not pretend this transformation is clean or risk-free. TeraWulf’s first quarter also included a net loss of $427.6 million, driven largely by non-cash warrant revaluation, stock-based compensation and impairment charges, a reminder that “AI pivot” has become a favored narrative for companies still carrying ugly balance sheets and capital-intensive legacy mining assets. In other words, AI is not rescuing miners from hard economics; it is giving them a more attractive story to tell while they try to refinance themselves into relevance.
The Bitcoin miner pivot is real, and slightly desperate
TeraWulf is not alone. Across the sector, miners are redirecting power from Bitcoin toward AI data centers because AI customers sign longer contracts, offer more predictable cash flow and care less about Bitcoin’s halving cycle than whether a site can deliver power on time.
Crypto.news has already tracked this broader shift. In one earlier story, Core Scientific sold $208.3 million worth of Bitcoin in the first quarter to fund its own AI data center buildout, with colocation revenue rising to $77.5 million while mining revenue fell to $30.1 million. In another crypto.news piece, Core outlined plans to convert its Pecos, Texas mining site into a 1.5 GW AI campus, repurposing about 300 MW previously used for Bitcoin mining.
The uncomfortable conclusion is that this is not a side strategy anymore. TeraWulf’s Kentucky deal shows that listed miners increasingly see hashpower as a transitional use case for energy-rich sites, not the endgame. Bitcoin may have built these companies, but AI compute is what they now believe can justify their valuations.
