CleanSpark’s latest numbers were ugly: revenue fell 30.5% year over year to $138 million for the three months ended June 30, and the Bitcoin miner posted a $239 million net loss, missing the $142.2 million revenue estimate tracked by Yahoo Finance.
- Revenue: $138 million, down 30.5% year over year
- Net loss: $239 million, or $0.89 per basic share
- AI/HPC push: a 20-year Sandersville lease could bring $6.6 billion in contracted revenue
- Bitcoin still bites: fair value losses on BTC holdings remain a major earnings swing factor
The company’s results are a reminder that Bitcoin miners do not get to coast on hype forever. When Bitcoin is strong, miners look like they have the Midas touch. When the market cools and accounting rules kick in, the financials can turn into a blunt lesson in volatility.
CleanSpark reported a net loss of $239 million, or $0.89 per basic share, versus net income of $257 million, or $0.90 per share, in the same quarter a year earlier. Revenue also dropped from $198 million in the prior-year period to $138 million.
That is a serious swing, not a rounding error. Investors have every reason to ask whether the business is being hit by weaker mining economics, the usual Bitcoin price whiplash, or both.
Why the quarter looked so rough
Part of the answer is simple: Bitcoin mining is a brutal business when margins tighten. Electricity is expensive, hardware ages fast, and block rewards are not exactly famous for their consistency. If the coin price does not cooperate, the whole model gets squeezed.
Then there is the accounting issue. Miners that hold Bitcoin on their balance sheets can take large fair value hits when BTC falls. A fair value loss is an accounting adjustment that reflects a change in the market value of those holdings. It is not the same thing as a cash loss, but it still hits reported earnings hard and makes quarterly results look far worse than the operating picture alone might suggest.
That is one of the joys of mining-company reporting: even if the rigs are humming, the income statement can still look like it got shoved down a staircase.
In CleanSpark’s fiscal second quarter ended March 31, the company said it held $925.2 million worth of Bitcoin and $260.3 million in cash. It also disclosed a $224.1 million fair value loss on its Bitcoin holdings, which accounted for nearly 60% of its total net loss in that period.
That number matters because it shows how much of the pain can come from mark-to-market accounting rather than pure operating failure. But it also shows the risk of keeping a large BTC treasury when the asset itself is volatile. Balance sheet strength and earnings volatility can coexist, and miners live right in that awkward middle.
The company also said in that quarter that its Bitcoin holdings increased 14% from a year earlier and average monthly hashrate climbed 18%. Hashrate is the amount of computing power devoted to mining, and higher hashrate generally means more capacity. It does not, however, magically cancel out weak pricing or power costs. Physics and economics remain rude to everyone.
For a useful comparison, see CleanSpark Reaches 10, 000 BTC Milestone: Impact on Bitcoin.
CleanSpark is trying to build a second engine
CleanSpark is not just trying to survive on block rewards. On July 14, the company signed a 20-year lease for a 175-megawatt data center at its Sandersville, Georgia campus with an undisclosed investment-grade global technology company.
The company said the deal could generate about $6.6 billion in contracted revenue over the initial lease term. That is a big headline number, but it needs to be read properly. Contracted revenue is not profit, not free cash flow, and not a guarantee that execution will be painless. A long lease can be valuable, but buildouts, power, cooling, financing, and operations still have to be handled without turning the whole thing into an expensive science fair project.
Still, the strategic logic is clear. CleanSpark owns or controls infrastructure that can support dense power loads, which is exactly what AI and high-performance computing, or HPC, need. These workloads use powerful computing systems for things like artificial intelligence and data processing. For miners with land, power access, and data-center footprints, that makes AI hosting and colocation a natural extension rather than a wild pivot.
During its May earnings release, chief executive Matt Schultz said the company planned to commercialize assets suitable for AI and HPC workloads while continuing to operate its Bitcoin mining business efficiently.
“CleanSpark has reported a $239 million quarterly loss as revenue fell 30.5% year over year, while its latest AI data center lease has added a long-term revenue stream outside Bitcoin mining.”
That is the tightrope the sector is now walking. Mining is still the core. But if the business is going to grow without depending entirely on the next Bitcoin spike, companies need to do more than stack ASICs and pray.
It also helps that CLSK stock has been in focus after CleanSpark signs $6.6B AI deal chatter hit the market, although traders love a shiny narrative almost as much as miners love cheap power.
The wider miner pivot is hard to ignore
CleanSpark is not alone in this shift. MARA reported a $1.3 billion first-quarter loss. TeraWulf Inc. Financial Results for Q1 2026: Net Loss and showed high-performance computing revenue exceeded Bitcoin mining revenue for the first time in the first quarter. Core Scientific posted a $347.2 million first-quarter loss while increasing colocation revenue as more capacity was allocated to AI infrastructure.
That broader pattern says a lot about where public miners are headed. Many are becoming power-and-data-center businesses with Bitcoin attached, rather than pure-play mining firms. Depending on your outlook, that is either smart diversification or a quiet admission that mining alone is too volatile to support a durable public-market valuation.
Both readings have merit.
The bullish case is that miners already own valuable infrastructure: land, power connections, cooling systems, and industrial-scale buildings. If those assets can serve AI customers, they can generate steadier, contract-based revenue and reduce dependence on Bitcoin’s mood swings.
The skeptical case is just as real. AI and HPC buildouts can be capital-intensive, customer concentration can be a problem, and a signed lease does not automatically mean healthy margins. Press releases are cheap. Executing large-scale infrastructure on time and on budget is the part where the adults show up.
Evergreen Language Module Overview is a reminder that even strong-looking operational updates can hide a lot of accounting noise, while Evergreen Language Module Details underline how much of the company’s pitch now depends on long-duration infrastructure revenue rather than hashprice alone.
CleanSpark’s 20-year lease is more credible than a lot of the vaporware nonsense that gets passed around in crypto circles, but the market still has to see what the economics look like in practice. A named customer would help. Better visibility on power costs, buildout spend, and timing would help more. Until then, the $6.6 billion figure should be treated as a long-term revenue projection, not a shortcut to instant profitability.
How the market reacted
CleanSpark shares fell 5.5% during Thursday’s trading session, then recovered about 3% in pre-market trading Friday to trade above $13.10. Google Finance data at the time showed a market capitalization of about $3.66 billion and a 52-week range of $8.00 to $23.61.
That range tells its own story. Miner stocks can move from “Bitcoin proxy” to “broken earnings machine” in a heartbeat, and then back again if the market decides the next narrative is AI, data centers, or some combination of the two. The volatility is not a bug. It is the product.
The more important question is whether CleanSpark is becoming something sturdier than a pure mining bet. The answer, for now, is yes and no. Bitcoin remains central, but the company is clearly trying to turn its infrastructure into a broader compute business. That is sensible. It is also a sign that the old miner model, by itself, is getting squeezed.
Bitcoin mining still has a place in the broader digital infrastructure stack. It secures the network, monetizes stranded or underused energy, and creates a native industry around hard assets rather than empty promises. But the public miners that survive the next few years will likely be the ones that can do more than mine coins and hope for price appreciation. They will need to operate like infrastructure companies, not lottery tickets with power bills.
For a broader context on miner economics, see Bitcoin Mining Difficulty Drops 10% in Rare Downward adjustment.
Key takeaways
-
Why did CleanSpark’s quarter look so bad?
Revenue fell to $138 million and the company posted a $239 million net loss. A large fair value loss on Bitcoin holdings was a major drag, which is common for miners that keep BTC on their balance sheets. -
Is CleanSpark dropping Bitcoin mining?
No. Bitcoin mining is still core, but the company is expanding into AI and HPC infrastructure to reduce reliance on mining-only revenue. -
What does the Sandersville lease mean?
CleanSpark says the 20-year lease could generate about $6.6 billion in contracted revenue. That is meaningful long-term top-line potential, but it is not the same as profit or free cash flow. -
Why are miners moving into AI and data centers?
Because AI and HPC contracts can provide steadier revenue than Bitcoin mining alone. That stability comes with real costs, but it is attractive in a volatile sector. -
Does this make CleanSpark healthier?
Potentially, yes. But the upside depends on execution, capital discipline, and whether the new revenue streams can offset the wild swings that come with Bitcoin exposure.
For further background on the company’s balance sheet strength and treasury build-up, see CleanSpark Reports Transformative FY 2025 Results.
And if you want to understand the AI customer side of the equation, CoreWeave is a useful reference point for how heavily infrastructure-heavy AI demand can reshape the economics of compute.
