The former Bitcoin miner is swapping pickaxes for GPUs, redirecting 28 MW of mining power toward AI workloads at its Michigan data center.
Hyperscale Data, the company trading under the ticker $GPUS, just signed a deal that could be worth more than $1.2 billion over the next decade. The Master Services Agreement, inked on June 24 with an unnamed California-based neocloud provider, commits the company to delivering 20 megawatts of AI compute capacity at its Michigan data center.
Here’s the thing: that 20 MW is just the opening act. The customer has options to scale up to 52 MW, with a further right to an additional 32 MW on top of that. If all expansion options get exercised, the total deal value could exceed $3 billion.
From mining rigs to AI racks #
Hyperscale Data is reallocating approximately 28 MW of power currently dedicated to Bitcoin mining operations at its Michigan campus and redirecting it toward AI and high-performance computing services.
The company plans to invest an estimated $100 to $120 million in capital expenditures to retrofit around 60,000 square feet of its data center for AI workloads. Bitcoin mining operations won’t disappear entirely. Hyperscale Data intends to maintain its mining activities at a separate site in Montana while the Michigan campus undergoes its transformation.
A campus with room to grow #
One of the more interesting details in this deal is the sheer amount of headroom at the Michigan facility. The campus has long-term capacity exceeding 300 MW. Even if the customer exercises all expansion options and takes the full 52 MW plus the additional 32 MW, that would still only represent roughly 28% of the site’s total potential.
The deal was not exactly a surprise to anyone paying close attention. Hyperscale Data disclosed that negotiations were at an advanced stage on June 15, nine days before the MSA was formally signed. That early disclosure triggered a significant market reaction on its own, with shares surging over 24% intraday when the official announcement dropped.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our