Hyperscale Data halts Michigan Bitcoin mining to repurpose 20MW for an AI customer
The contract is framed at $1.2B+ over 20 years with extensions, and $3B+ if a 32MW option is exercised.
Hyperscale Data shut down Bitcoin mining at its Michigan data center on Sept. 1 to prepare the facility for an AI cloud-computing customer under a long-term capacity agreement. The company framed the deal at more than $1.2 billion over 20 years with extensions, with upside above $3 billion if the customer adds 32 megawatts within two years.
Hyperscale Data stopped Bitcoin mining at its Michigan data center on Sept. 1 and is repurposing the site for an AI customer, shifting the facility’s power budget away from hashrate and toward contracted data center capacity. The agreement covers 20 megawatts (MW) for an unnamed California-based provider of cloud computing services for AI, with the company positioning the move as an operational retooling rather than a parallel build.
MW capacity is the hard ceiling here. It is the amount of electrical power the site can deliver to compute equipment, which caps how much work the facility can run whether that is mining rigs or AI servers.
CEO William Horne tied the shutdown directly to execution on the AI buildout. “The immediate shutdown of the Bitcoin mining operations allows our team to focus the Facility's power, infrastructure and resources in preparing the Facility for its usage by our Customer,” he said.
The economics being marketed to investors are explicitly option-driven. Hyperscale Data said the contract could generate more than $1.2 billion over 20 years if the customer exercises both five-year extensions to an initial 10-year term, meaning the headline number is not the base term but the full extension path.
Hyperscale Data also said it plans to sell its mining servers and expects gains from those sales, framing the hardware exit as a balance-sheet lever during the transition. The company did not announce a start date for AI operations at the Michigan facility, leaving the timing of any revenue ramp and the duration of the “dark” period between mining shutdown and AI commissioning unclear.
Catalysts That Decide Whether This Is a $1.2B Deal or a $3B+ Deal
The contract’s top-line range is a ladder of customer choices, not a single guaranteed revenue stream. The first rung is simply getting the 20MW live and billable, and the disclosure so far does not include a commissioning date or phased milestones that would let traders map revenue recognition to a calendar.
The second rung is term length. The more-than-$1.2-billion figure depends on two five-year extensions, which are optional add-on periods that extend the initial contract term if the customer chooses to continue. Without those extensions, the deal is a 10-year capacity agreement with a smaller total, even if the facility is fully utilized.
The third rung is the expansion option. Hyperscale Data said the customer has an option to add 32MW within the first two years, and that exercising that option plus both five-year extensions could bring total contract revenue above $3 billion. That makes the “$3B+” case a compound scenario that requires both near-term expansion appetite and long-term renewal behavior.
Two other details will decide how tradable the story becomes. One is counterparty clarity: the customer is unnamed beyond being a California-based AI cloud provider, so there is no way yet to underwrite credit, concentration risk, or whether the buyer is a hyperscaler-adjacent aggregator versus a thinner-margin reseller. The other is the capex and retrofit path implied by the pivot, where the sector’s recent history has shown that repurposing can come with real accounting pain. IREN’s quarterly results released last month showed AI cloud revenue surpassing Bitcoin mining revenue for the first time, but it also recorded a $450.4 million asset write-down, mostly tied to retired mining equipment.
My Read: Miner-to-AI Pivots Are About Contracted Power, Not Hashrate Headlines
The part that decides this trade is not the narrative that “miners are becoming AI companies,” it is whether the 20MW becomes contracted, delivered power with a start date and a verifiable counterparty. Hyperscale Data has already made the irreversible move by halting mining at the Michigan site, but the disclosure still leaves the most important timing variable blank, and that is where these pivots tend to slip.
The threshold that matters is whether the company can turn the unnamed customer and the commissioning timeline into something the market can model, because the $1.2B+ and $3B+ figures only exist if the customer renews twice and expands within two years. If those options get exercised and the facility actually ramps, the story stops being a hashrate headline and starts being a contracted-power re-rating story.
