Bitcoin miners pursuing artificial intelligence and high-performance computing revenue are facing a costly transition: MARA Holdings and CleanSpark both reported steep year-on-year revenue declines and large net losses in their latest quarters while committing more power, land, and data-center capacity to non-mining customers.
MARA reported second-quarter revenue of $174.9 million, down 27% from $238.5 million a year earlier. CleanSpark’s revenue for the fiscal third quarter ended June 30 fell 30.5% to $138 million from $198.6 million in the prior-year period. Both companies expanded their operating footprints during the quarter, but much of that capacity remains tied to projects that require construction, grid work, customer deployment, or regulatory approvals before they can produce material AI-related revenue.
The results show the financial tension in the mining-to-compute conversion. Bitcoin miners can deploy power quickly to their own fleets, yet converting energy-rich sites into facilities suitable for high-performance computing customers requires specialized buildings, cooling systems, fiber connections and long-term customer commitments. Those projects can consume capital well before servers generate lease income.
MARA shares fell more than 5% to $10.67 following the earnings release and updated operating figures, while CleanSpark shares declined more than 6% to $12.69. The market reaction reflected pressure on companies whose valuations increasingly depend on their ability to turn large power portfolios into operating data-center businesses rather than simply announce future capacity.
MARA’s losses widened despite higher hashrate
MARA posted a net loss of $611.3 million, or $1.60 per diluted share, compared with net income of $808.2 million a year earlier. The company recorded a $343 million fair-value loss on digital assets during the quarter, illustrating how large bitcoin holdings can amplify reported earnings swings even when mining operations improve.
Adjusted EBITDA, a measure that excludes items such as interest, taxes, depreciation and certain non-cash charges, moved to a loss of $360.9 million from a $1.2 billion profit in the comparable quarter a year earlier, according to MARA.
The company mined 2,422 BTC during the quarter at an average price of roughly $71,325 and sold 2,213 BTC at an average price of $73,078. Its energized hashrate, a measure of mining machines actively operating, rose 22% year over year to 70.3 exahashes per second. MARA also said cost per petahash per day improved 4% to $27.70.
Those operating gains did not prevent a decline in MARA’s bitcoin treasury. Its holdings fell 29% to 35,577 BTC, valued at about $2.1 billion based on the company’s reported figures. Bitcoin Treasuries ranked MARA fourth among public companies by bitcoin holdings. The size of that reserve gives MARA substantial exposure to bitcoin’s price movements, while also providing a potential source of liquidity for its infrastructure plans.
CleanSpark reports a similar earnings reversal
CleanSpark reported a net loss of $239.8 million, or $0.89 per basic share, after recording net income of $257.4 million, or $0.90 per share, in the prior-year quarter. Its adjusted EBITDA shifted to a $113 million loss from a $377.7 million profit a year earlier.
The company also reported a $116.3 million fair-value loss on bitcoin. As of June 30, CleanSpark held $202.6 million in cash and bitcoin valued at $814.9 million. Bitcoin Treasuries placed its 13,924 BTC reserve eleventh among public companies.
CleanSpark listed total assets of $2.7 billion, long-term debt of $1.8 billion and working capital of $761 million. The balance-sheet figures place the company in a position where the pace of infrastructure spending, debt servicing and bitcoin-market volatility will remain central to how traders assess its conversion strategy.
CleanSpark said it controls more than 1.8 gigawatts of power, land and data-center capacity. That figure gives the company a sizeable base for prospective hosting or high-performance computing development, but control of capacity is different from fully commissioned, revenue-generating data-center operations.
The company previously disclosed a 20-year, $6.6 billion triple-net lease in Sandersville with a tenant described as “high investment-grade.” Under a triple-net structure, the tenant generally bears operating expenses such as taxes, insurance and maintenance. The agreement could provide a more predictable revenue model than bitcoin mining once the underlying facility is delivered and operational.
Power portfolios become the central metric
MARA is pursuing a similar expansion through acquisitions and site development. The company said it is working to close its Long Ridge acquisition, which it expects to provide immediate positive EBITDA and increase AI capacity at its Hannibal campus. The transaction remains subject to approval from the Federal Energy Regulatory Commission.
MARA also acquired a site in Matagorda County, Texas, adding 2 gigawatts of power capacity. The company said the acquisition could bring its overall power portfolio toward 4.8 gigawatts. Such capacity gives MARA more options to allocate electricity between bitcoin mining, AI computing and other data-center uses, depending on economics and customer demand.
The near-term challenge is execution rather than access to energy. Bitcoin mining equipment can be installed in modular facilities and operated under variable power conditions. High-performance computing customers generally require more stringent uptime standards, network connectivity, cooling infrastructure and contractual delivery schedules. A large power number can therefore support a long-term strategy without immediately changing quarterly revenue.
Both companies entered the period with meaningful bitcoin reserves, expanding power access and mining operations that remain large by industry standards. Their latest earnings indicate that traders are likely to focus more closely on whether announced sites move from land and power rights into commissioned facilities producing contracted data-center cash flow.
Explore how mining fits the AI boom with Toobit’s AI–blockchain insights for smarter, future-proof crypto infrastructure decisions.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
