Hut 8 secures $9.8B AI lease to fill 1 GW Texas campus

Hut 8’s 1‑gigawatt Beacon Point campus moving to full commercialisation is another data point in how quickly AI compute is industrialising. A single site now has nearly a terawatt‑hour equivalent of contracted AI capacity with a 15‑year revenue line that rivals many hyperscalers’ regional footprints. This is what the post‑GPU‑shortage world looks like: capital markets are underwriting multi‑billion‑dollar, multi‑decade AI power plants built around long‑term offtake agreements.

For the race to AGI, the significance is not in Hut 8 specifically but in what it says about bottlenecks. As former crypto miners repurpose power‑dense sites into AI campuses, compute availability is migrating from being a theoretical constraint to a financial one. If enough gigawatt‑scale campuses lock in investment‑grade tenants, the limiting factor for training and running frontier‑class models will be less about electricity and rack space and more about talent, algorithms and safety assurance.

Strategically, this kind of contract also resets the bargaining power between AI labs and infrastructure providers. When a single, unnamed tenant can commit to hundreds of megawatts, it suggests hyperscalers and frontier labs are willing to sign utility‑like offtake deals to secure capacity. That may favour players with deep balance sheets and access to cheap power, and will squeeze smaller model labs who can’t guarantee similar multi‑decade commitments.