Hyperscale Data (NYSE American: GPUS) has pulled the plug on Bitcoin mining at its Michigan facility, completing a full transition to AI computing services as of September 1, 2026. The reward for this strategic reinvention?
Shares trading at all-time lows. The company signed a master services agreement with an unnamed California-based neocloud provider to supply AI compute infrastructure, starting with an initial capacity of 20 MW. The deal is projected to generate over $1.2B in revenue across its initial 10-year term, with expansion options that could push the total past $3B. From hash rates to neural networks
Hyperscale’s Michigan campus spans 617,000 square feet across roughly 83 acres. The site has existing infrastructure for up to 340 MW of capacity, but the current AI deal only taps about 20% of that potential. That leaves significant room for additional customers as demand for high-performance computing continues its upward trajectory. A $1.2B floor over 10 years translates to roughly $120M in annual revenue before any expansion kicks in.
An additional 32 MW of capacity could be brought online, which the company says would boost the total contract value beyond $3B. Why the stock keeps falling
GPUS shares have been in freefall, hitting all-time lows despite the headline-grabbing deal size. Hyperscale Data has put shareholders through a series of reverse stock splits, which have eroded confidence in management’s ability to create organic value.
The neocloud provider at the other end of this MSA remains unnamed, which adds another layer of uncertainty. A broader industry pattern
Hyperscale Data hasn’t entirely abandoned crypto. The company reportedly continues to explore Bitcoin mining operations in other regions, including Montana. The Michigan facility’s sheer scale gives it a theoretical advantage. With 340 MW of potential capacity and only 20% currently committed, the campus could accommodate a significant ramp in AI workloads without requiring major new infrastructure investment. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
