Hyperscale Data has switched off the last of its Bitcoin miners in Michigan, formally ending the company’s run as a public-market Bitcoin producer as it converts the site into an artificial intelligence data center under a contract expected to be worth approximately 1.2 billion USD.
The company said Wednesday that all Bitcoin miners at the Michigan facility were powered down following an inspection by its customer, an unidentified California-based neocloud provider. Hyperscale said it intends to sell the associated mining equipment, closing the chapter on a mining operation that was still its operational identity a year ago.
The AI customer has contracted for 20 megawatts of computing capacity under a 10-year master services agreement carrying two optional five-year extensions. Over the maximum 20-year term, the agreement may generate more than 1.2 billion USD in revenue. An additional 32-megawatt option could push potential revenue above 3 billion USD, and the site is expected to ultimately support 340 megawatts of capacity.
A treasury drained to fund the pivot
The mining shutdown is the visible half of a broader transformation that has been funded, in large part, by selling down the company’s Bitcoin treasury.
On July 30, Hyperscale Data held approximately 1,006 Bitcoin and had sold 100 BTC while arranging a BTC-backed credit facility to help fund the Michigan campus buildout. By Tuesday, the picture had changed drastically: the company disclosed that it sold approximately 65 BTC for approximately 5.1 million USD during the week ending August 30, with proceeds directed toward additional capital for the Michigan development.
BitcoinTreasuries.NET now lists Hyperscale Data as holding 215 BTC, worth approximately 16.7 million USD. That represents a 79 percent decline from the holdings cited in July, and it drops the company to 84th place among the public companies tracked by the platform. For a firm that once marketed itself partly on its Bitcoin accumulation strategy, the drawdown marks a wholesale liquidation of that position to bankroll an AI infrastructure bet.
The company has cautioned that its expansion plans remain preliminary and subject to financing, approvals and other risks. The 1.2-billion-USD estimate requires the customer to exercise both five-year extension options, and the 3-billion-USD projection additionally depends on the customer taking up the 32 megawatts of extra capacity. In other words, the headline revenue figures describe the most extended scenario, not a booked backlog.
Record low despite the AI pivot
The market has not rewarded the strategy so far. According to Yahoo Finance data, Hyperscale Data shares closed at 0.1984 USD on Wednesday, down about 17 percent, after touching an intraday low of 0.1932 USD. The close marked a split-adjusted record low for the NYSE American-listed stock.
The decline came shortly after the company completed a one-for-five reverse stock split, with shares trading on a split-adjusted basis from August 25, according to a filing with the U.S. Securities and Exchange Commission. Reverse splits are typically used to maintain listing compliance, and the subsequent slide to record lows suggests the measure has not restored investor confidence.
The irony is sharp: the miners were switched off to make room for a customer paying contracted rates for AI compute, a business model the market claims to want exposure to, yet the stock reacted by falling further.
The miner-to-AI conversion wave
Hyperscale Data is not alone in this migration. Former Bitcoin mining sites across North America have become the fastest-growing source of converted capacity for AI and high-performance computing, because they already control the hardest part of the data center equation: powered land, grid interconnections and cooling infrastructure. Neocloud providers, which lease GPU capacity to AI developers, have been signing long-term contracts to lock in that electrical capacity.
The economics explain the rush. A 10-year master services agreement with a creditworthy counterparty produces contracted, utility-like revenue, while Bitcoin mining revenue floats with hash price, difficulty and the coin’s market price. For a company with Hyperscale’s balance sheet pressures, the certainty of a 1.2-billion-USD potential contract stream evidently outweighed the optionality of holding miners and a four-figure Bitcoin stack.
But the transition carries real costs, and Hyperscale’s numbers illustrate them. The company is selling mining equipment into what is likely a soft secondary market for older-generation machines, it liquidated 79 percent of its Bitcoin treasury largely to fund construction, and its equity has hit all-time lows even as the AI narrative peaks. Conversion is capital-intensive, and the payoff depends on customers exercising options that remain exactly that: optional.
What it signals for miners and Bitcoin treasuries
For the mining sector, the Michigan shutdown is another data point in a consolidation story: smaller, leveraged operators with powered sites are exiting proof-of-work and monetizing their megawatts elsewhere. For corporate Bitcoin treasury watchers, it is a reminder that treasuries labeled as strategic can turn out to be transitional, sold down when capital needs collide with a liquidity window.
The remaining question for Hyperscale is execution. The miners are off, the land is cleared for AI racks, and the customer is under contract for 20 megawatts. Whether the 340-megawatt vision, the 3-billion-USD upside and a rebuilt equity story follow depends on financing and options that have not yet been exercised. For now, one fewer public company mines Bitcoin, and one more data center campus belongs to artificial intelligence.
sold 79% of the BTC treasury to become an AI landlord for one unnamed neocloud. if that customer walks, this is a shell with racks in michigan
single unnamed customer on a 10 year MSA is a counterparty bet dressed up as a pivot. at least mining revenue diversified across the whole network
if the neocloud were a blue chip theyd have named it. unnamed customer means nda or they dont want you googling their balance sheet
the inspection already passed which is something, but an unnamed counterparty on a billion dollar headline is still marketing until the 10-K names them
mining revenue diversified across the network but hashprice still dictated everything. at least the MSA has contracted rates, thats the whole trade
the 1.2 billion is also the stretched scenario, you need both five year extensions plus the 32 megawatt option. booked revenue today is far smaller
and 5.1 million from 65 BTC covers maybe a month of capex at 20MW. runway math gets ugly fast if the michigan buildout slips
Exactly, and that 5.1 million assumes decent fills on all 65 BTC. A thin treasury plus contingent capex is a rough combination.
and they funded the bridge by selling 65 BTC at whatever the market gave them that week. the capex gap and the treasury dump are the same bad trade
yep, and the 1.2B needs both five year extensions exercised. base case is closer to 600M over the first ten years
600M over ten years still beats mining margins at this hashprice. the ugly part is funding the bridge with 79 pct of the treasury, not the MSA itself
one unnamed neocloud on a 10yr MSA after selling 79% of the treasury. if they had named aws this would be a very different stock
every miner pivot story quotes the max contract, nobody quotes who signs it. if the first 20MW underdelivers that 32MW option dies quietly and the 3B number goes with it
84th largest public BTC holder and dropping. every small miner is making this exact pivot at the same time, ai hosting margins will compress fast
the whole sector is racing down that holder list at once. hpc margins look great until every miner with racks is chasing the same ten neoclouds
converted datacenter supply is exploding while neocloud demand sits with a handful of buyers. someone is ending up holding depreciating racks
every quarter another miner announces the same ai hosting pivot. all that capacity landing at once is how 2026 margins end up looking like 2018 mining
the neocloud inspection clause is the detail everyone skips. one unnamed california buyer can just not show up, the miners are already on the used market, and 215 BTC is all thats left of the old plan
record low while announcing the pivot is the market doing math on the 1.2b being a stretch case. needs both extensions plus the 32mw option lol