CleanSpark reports 593 BTC for August: reading the hashrate gap
Source report: 2026-09-08 · Editorial analysis published: 2026-09-10
CleanSpark’s unaudited August update reports 593 BTC, 50 EH/s operational hashrate and 38.3 EH/s average operating hashrate. These measures answer different questions.
Analysis and practical implications
This section is our analysis and illustrative calculations, separate from the source report.
The figures describe different boundaries
CleanSpark published its unaudited August operating update on September 8, reporting 593 BTC produced, operational hashrate of 50 EH/s and average operating hashrate of 38.3 EH/s. Its operational measure refers to a concurrently achieved level for installed, functioning equipment. An average describes activity over a period. Neither should silently replace the other when assessing what the installation actually delivered during the month.
The distinction is relevant beyond a single company. Product specifications, installed capacity, a demonstrated peak and a period average can all be valid numbers while measuring different things. ASIC.tools interprets the announcement as a reason to preserve those definitions in an operating report. The discussion below is our analysis of measurement, not a diagnosis of why this particular fleet's two figures differ.
Do not turn a ratio into an uptime claim
Dividing an average by a higher reference produces a mathematical ratio. It does not automatically produce the percentage of machines that were available, the percentage of hours that were productive or a measure of hardware reliability. Those conclusions require compatible definitions and the underlying records. A fleet can change its operating profile or composition during a period, making a single aggregate ratio an incomplete account of how it worked.
Keep the original labels beside the numbers in any comparison. If a calculation uses a reference level achieved at a different point, state that limitation. Do not infer a specific amount of downtime, curtailment or equipment failure from the difference alone. Several operating patterns can lead to similar averages. An explanation should be supported by recorded events rather than chosen because it sounds plausible beside the headline figures.
Reconcile production and operating conditions
To connect capacity with coin production, align accepted work with its observation interval and the settlement records of the relevant pool. Local hashrate displays are useful for diagnosing machines, but they are not substitutes for the work that was accepted and credited. Keep rejected work and connection interruptions visible. This makes it possible to distinguish a measurement difference from a reduction in productive delivery.
Energy should use the same interval and a clearly stated boundary. The electrical draw of the miners, the draw of the whole installation and the amount billed by a supplier can differ because of auxiliary equipment and the billing arrangement. A comparison of monthly production with an instantaneous power observation will not capture that relationship. Record measured energy and the associated invoice instead of reconstructing the whole month from a single favourable operating point.
Use the distinction in your own catalogue
For an individual operator, the catalogue specification is a starting reference. Save a separate record of the installed configuration and the performance achieved under the site's own conditions. Firmware version, selected profile and equipment changes help explain differences over time. Updating that record does not require changing the manufacturer's specification; the two records serve different purposes and should remain distinguishable.
A useful dashboard can therefore show the reference specification, demonstrated capacity, period-average accepted work and actual production together, each with a definition and timestamp. This structure makes comparisons more transparent without pretending that one metric explains every operational outcome. The practical takeaway from the August update is to ask which boundary a number describes before using it for a purchasing decision, a fleet assessment or a profitability model. For a reusable dashboard, retain the source label, unit, reporting interval and definition next to each number. A historical concurrent maximum should remain distinguishable from a monthly average even when both use EH/s. When comparing months, check whether the definitions stayed the same before interpreting a change as improved or worse operations. If a definition changes, annotate the break in the series. This is a method for reading company disclosures, not a new finding about the causes of CleanSpark’s August performance.
Source: CleanSpark ↗
Mining calculator ↗
