CleanSpark produced 593 BTC and sold 821 in August, and ended the month with fewer BTC than it started with. Production improved a bit. Treasury balance fell more than that. In mining, that is often where the real story lives.
- 593 BTC mined in August, up from 586 in July
- 821 BTC sold, more than new production
- 13, 703 BTC held at month-end, with 3, 951 tied to derivatives
- Mining economics improved, but full profitability is still unproven
CleanSpark’s Sept. 8 operating update shows a miner that is still producing, still building, and still leaning on structured sales and hedges instead of a simple “mine and stack” playbook. The company produced 593 BTC in August 2026, up from 586 BTC in July, while average operating hashrate came in at 38.3 exahashes per second, slightly below July’s 38.6 EH/s.
The headline numbers are simple enough. The meaning behind them is messier.
CleanSpark sold 821 BTC in August, including 77 BTC sold at spot prices, 500 BTC sold pursuant to call exercises, and 244 BTC related to a delta-neutral basis trade. It ended the month with 13, 703 BTC, down from 13, 931 BTC on July 31. Net result: the treasury shrank by 228 BTC even though production rose.
That does not automatically signal trouble. It does show that CleanSpark is managing Bitcoin like a corporate treasury asset, not a sacred pile of digital gold that must never leave the wallet. The delta-neutral basis trade is a hedging strategy that usually offsets Bitcoin price risk with derivatives while trying to capture the basis between spot and futures. Useful? Sure. Simple? Not even close.
The more important detail is that 3, 951 BTC of CleanSpark’s holdings were posted as collateral or recorded as a receivable, all tied to derivative transactions. That means a chunk of the reported treasury is economically encumbered. It is not the same thing as free-and-clear coins sitting there ready for a rainy day or a buyback fantasy.
CleanSpark reported peak fleet efficiency of 16.07 joules per terahash, deployed 201, 269 machines as of Aug. 31, and said its operational hashrate remained at 50 EH/s. It also reported maximum concurrent power use of 808 megawatts against 1.8 gigawatts of contracted capacity. These are large-scale industrial figures, not garage-miner cosplay.
Joules per terahash measures how much energy a mining machine uses to perform a unit of hashing work. Lower is better. A lower number usually means a miner can survive a rough market more comfortably because it squeezes more work out of each watt.
That said, CleanSpark’s 16.07 J/TH figure is a peak measurement, not a fleet-wide average. That distinction matters. Peak efficiency tells you the best-performing installed hardware is strong. It does not tell you whether the entire fleet is equally efficient, or whether the business is profitable after labor, maintenance, financing, overhead, and accounting drag.
The company itself said rising Bitcoin prices improved mining revenue during the second half of August. That lines up with the broader point: hashprice is the revenue miners receive per unit of computing power, and it generally improves when Bitcoin prices rise or network conditions get friendlier. In plain English, the market gave miners a better hand late in the month.
CleanSpark’s August production also included a peak single-day haul of 20.40 BTC and an average of 19.12 BTC per day. Year to date through Aug. 31, it had produced 4, 903 BTC. Those are solid output numbers, especially in a sector where weak operators get crushed by electricity costs and obsolete hardware.
But production is not the same thing as profitability. That is where a lot of mining coverage gets lazy and starts hallucinating strength where there may only be survivability.
The broader accounting picture matters. In its quarterly filing, CleanSpark reported $138 million in Bitcoin mining revenue, $85.5 million in cost of revenue, a company-wide net loss of $239 million, and a $116.3 million Bitcoin fair-value loss. That does not mean the miners were losing money on every coin they produced. It does mean the corporate machine can still post a brutal bottom line even when the rigs are doing their job.
Bitcoin fair-value loss is an accounting loss that comes from changes in the value of a company’s Bitcoin holdings. It is not the same thing as a realized cash loss, but it still hits reported earnings. For miners, that distinction matters because the operating business and the balance sheet can tell very different stories.
The safest read is this: August likely brought better mining economics, and CleanSpark’s more efficient machines probably helped. But the available information does not prove the entire operation was net profitable once all costs are counted. Peak efficiency and stronger BTC prices are helpful. They are not magic. Unfortunately for promotional decks everywhere, electricity bills still exist.
CleanSpark’s infrastructure buildout remains a separate but related story. The company highlighted ongoing construction at its Sandersville, Georgia data center campus, which it associates with $6.6 billion in contracted revenue. That is a serious figure, but it should be handled with care. Contracted revenue is not the same as revenue already recognized. A big number on a slide is not cash in the bank.
Bitcoin miners pivoting to AI data centers has become a familiar theme across the industry, with miners chasing higher-margin compute deals while the pure mining business gets squeezed. The pitch sounds smart. The risk is that the network’s security budget gets treated like a side quest instead of the main event.
Texas remains the other major front.
ERCOT requests suspension of Batch Zero deadlines as the grid operator keeps sorting through industrial load interconnection. CleanSpark said ERCOT issued conditional batch-zero classifications covering 585 megawatts of contracted baseload capacity and 300 megawatts of studied load capacity. ERCOT, the Electric Reliability Council of Texas, runs the state’s grid and sits at the center of a lot of industrial mining expansion. Here, though, conditional is the key word. It means progress, not full approval.
Batch Zero classifications are part of the review process for large loads seeking grid interconnection. In plain English: the projects are being processed, not fully waved through. CleanSpark said it will continue working with the Public Utility Commission of Texas as that process advances.
That matters because power access is the entire game in large-scale mining. Bitcoin may be decentralized, but mining still runs on very centralized realities: megawatts, interconnection queues, regulators, and utility politics. The romantic version is “secure the network.” The actual version is “please let us plug in the transformers without setting off a bureaucratic chain reaction.”
CleanSpark’s sales breakdown also hints at a more sophisticated treasury posture than simple liquidation. Selling 500 BTC through call exercises and 244 BTC via a delta-neutral basis trade suggests the company is actively managing exposure and monetization, not just dumping coins to cover the light bill. That can be smart. It can also make the balance sheet harder to read from the outside.
The average sale price of $65, 420 per BTC is useful context. It suggests the company monetized its coins at a relatively strong level during the month. That helps. But good execution on sales does not automatically translate into strong all-in economics if overhead and financing costs are chewing through the rest of the margin.
In other words, CleanSpark had a better August than a lot of weak miners would envy. That still leaves a very reasonable question: did the whole business actually make money?
Answer: there is not enough here to say yes.
Key questions and takeaways
-
Did CleanSpark mine more Bitcoin in August than in July?
Yes. It produced 593 BTC in August, up from 586 BTC in July, even though average operating hashrate edged lower. -
Why did CleanSpark’s BTC holdings fall?
Because it sold 821 BTC while producing 593 BTC. That left it with 228 fewer BTC at month-end than it started with. -
Are all 13, 703 BTC freely available?
No. CleanSpark said 3, 951 BTC were posted as collateral or recorded as a receivable in derivative transactions, so that portion is not fully free for treasury use. -
Does the efficiency data prove the whole fleet was profitable?
No. The 16.07 J/TH figure is a peak measurement, not an average for every machine, and it does not prove full net profitability after all costs. -
What does the ERCOT update mean for Texas operations?
It means CleanSpark made progress, but only conditionally. The classifications are not full approval, so the Texas sites still have regulatory steps to clear.
CleanSpark is looking more like a hybrid industrial platform than a pure Bitcoin miner: part producer, part treasury manager, part infrastructure developer, part power-market participant. That may be the right model for the modern mining era. It also means the cleanest-sounding numbers are often the least informative ones.
Bitcoin mining rewards efficiency, but the winners now need more than good machines. They need cheap power, disciplined treasury management, and enough financial engineering to survive the ugly parts. The hash rate may be decentralized. The spreadsheets are not.
Further reading
A few extra resources for miners, treasury-watchers, and anyone tracking where the real margins are hiding.
