The corporate is making a wager a $1.2 billion AI compute deal will succeed where Bitcoin mining couldn’t, but investors are seemingly to be not procuring for it yet
Hyperscale Knowledge (NYSE American: GPUS) has pulled the lumber on Bitcoin mining at its Michigan facility, winding up a stout transition to AI computing products and companies as of September 1, 2026. The reward for this strategic reinvention? Shares trading at all-time lows.
The corporate signed a grasp products and companies settlement with an unnamed California-essentially based neocloud provider to present AI compute infrastructure, starting with an initial capability of 20 MW. The deal is projected to generate over $1.2B in earnings across its initial 10-year term, with expansion alternate solutions that could well also push the total previous $3B.
From hash rates to neural networks
Hyperscale’s Michigan campus spans 617,000 sq. feet across roughly 83 acres. The positioning has existing infrastructure for as a lot as 340 MW of capability, but the present AI deal only taps about 20% of that doubtless. That leaves fundamental room for additonal customers as request of for prime-performance computing continues its upward trajectory.
A $1.2B ground over 10 years interprets to roughly $120M in annual earnings sooner than any expansion kicks in. A further 32 MW of capability will most definitely be brought on-line, which the corporate says would boost the total contract price previous $3B.
Why the stock keeps falling
GPUS shares were in freefall, hitting all-time lows despite the headline-grabbing deal measurement. Hyperscale Knowledge has set shareholders through a series of reverse stock splits, which cling eroded self perception in administration’s capacity to develop natural price.
The neocloud provider at the many halt of this MSA remains unnamed, which provides one more layer of uncertainty.
A broader switch sample
Hyperscale Knowledge hasn’t solely abandoned crypto. The corporate reportedly continues to explore Bitcoin mining operations in various regions, collectively with Montana.
The Michigan facility’s sheer scale affords it a theoretical advantage. With 340 MW of doubtless capability and only 20% right now dedicated, the campus could well also accommodate a fundamental ramp in AI workloads with out requiring main unusual infrastructure funding.
Disclosure: This text was edited by Editorial Team. For more info on how we develop and review lisp material, glimpse our Editorial Policy.
