Summary
- Kentucky regulators have approved service for up to 482MW at TeraWulf's Justified Data Campus.
- The agreement allocates transmission, market, delivery, and customer-specific infrastructure costs to the large-load customer.
- The structure provides a useful comparator as European regulators confront who should fund grid reinforcement for data-centre growth.
Kentucky regulators have approved an electricity-service agreement covering up to 482MW for TeraWulf’s proposed Justified Data Campus in Hancock County, attaching a defined power-cost structure to one of the latest large US AI-infrastructure developments.
The Kentucky Public Service Commission approved the agreement between TeraWulf, Kenergy, and Big Rivers Electric Corporation on 21 August. The arrangement allows service to begin under the agreed framework while requiring regulatory approval for subsequent amendments.
The campus is being developed on the site of a former aluminium operation and is expected to make use of an existing 482MW grid connection. TeraWulf has said the first data-centre capacity could enter service in 2027.
The agreement is notable for more than its size. TeraWulf is responsible for power-market, transmission, delivery, and other costs attributable to its load, alongside customer-specific infrastructure and financial security arrangements.
That reflects growing concern among regulators and utilities that very large data-centre connections should not transfer disproportionate network costs or demand risk to existing electricity customers. A 482MW campus is comparable to the load of a substantial industrial complex, and its arrival can require dedicated network equipment, contracted generation, transmission capacity, and long-term planning changes.
Regulatory filings estimate several billion dollars of investment by TeraWulf and additional investment from tenants, although those figures remain project forecasts rather than completed expenditure. The load is expected to ramp rather than appear at full scale immediately.
The structure has a direct European parallel. DataCentral reported this month on Virginia’s decision to allocate dedicated transmission costs associated with new data-centre loads to developers. European grid bodies are facing a similar problem as connection queues fill with projects whose combined requested capacity can greatly exceed the likely build-out.
Assigning customer-specific costs can protect other consumers, but it does not solve every grid issue. Transmission reinforcement still takes time, generators need adequate capacity, and utilities must decide how much weight to give speculative or phased loads when planning future infrastructure.
The Kentucky project benefits from something increasingly valuable in global data-centre development: an industrial site with a large existing electrical connection. Reusing former heavy-industrial infrastructure can shorten parts of the power-development programme compared with creating a new connection of similar scale on undeveloped land.
It also illustrates why former industrial and energy sites are attracting AI infrastructure. Large land parcels, established substations, transmission corridors, and historical utility loads can make them easier to adapt to hundreds of megawatts of compute than conventional commercial property.
The Kentucky commission has nevertheless retained oversight. TeraWulf is expected to provide further information including noise and air modelling as the project develops, while any material changes to the service arrangement remain subject to regulatory scrutiny.
For European markets confronting much larger connection queues than operating data-centre loads, the useful comparison is the contract structure rather than the geography: secure the power, identify the infrastructure costs, and establish who carries the risk before hundreds of megawatts are allowed to reshape the network.
