Canaan reports 44 BTC for September and full operation of Nordic heating project
Source report: 2026-10-15 · Editorial analysis published: 2026-10-16
Canaan’s October 15 update separates September mining and treasury data from a newly operational heating project. Its theoretical operating-hashrate definitions and paused Ethiopia operations require care.
Analysis and practical implications
This section is our analysis and illustrative calculations, separate from the source report.
September results and an October update
Canaan released its unaudited September mining update on October 15, reporting 44 BTC mined and month-end holdings of 1,875 BTC. It also says its 8 MW Nordic district-heating project is fully operational, delivering recovered computing heat to approximately 2,800 homes. The Canadian greenhouse project remains planned deployment pending site inspections. Those are different project statuses, rather than two completed heat-reuse installations.
The dates have distinct roles. September production and September 30 inventory describe a completed reporting period, while the October release provides later project and treasury information. Neither a month-end snapshot nor an October announcement should be relabelled as the current minute’s operating measurement. Keeping the period alongside each number avoids combining historical mining output with a later asset transaction or a newly confirmed project milestone.
Operating hashrate is not a live pool measurement
The release gives non-joint-venture installed and operating hashrate as 10 EH/s each. Its non-JV operating definition is theoretical capacity from energized machines, including units that may temporarily be offline. This is not the same as accepted work measured continuously by a pool. An energized-capacity label cannot establish that every miner contributed throughout September or that a stated month-end rate was the monthly average.
There is a particularly important qualification: Canaan says mining operations in Ethiopia are paused, with that hashrate retained in installed capacity. Its regional table nevertheless lists Ethiopia at 4.96 EH/s in both operating and installed columns. We retain the reported labels and the pause together rather than resolving them into an invented live rate. This prevents a nominal total from being presented as verified simultaneous production across every location.
Joint ventures have a separate reporting boundary
Joint-venture capacity is shown separately at 4.87 EH/s installed and 4.18 EH/s operating. The relevant facilities are Alborz, Bear and Chief Mountain in West Texas, where Canaan states a 49% stake. The release excludes JV metrics from its Bitcoin production and average all-in power-cost calculations. The facility-level capacity numbers therefore should not be silently treated as wholly owned output behind the reported 44 BTC.
Our arithmetic check, 4.18 divided by 4.87, is approximately 85.83%. That is a ratio of two reported snapshot categories, not a measured monthly uptime percentage, a revenue share or a guarantee. Likewise, the reported global 14.18 EH/s operating and 14.87 EH/s installed totals combine categories whose scope still matters. They are not a sufficient denominator for estimating the yield of a particular ASIC or allocating the company’s BTC output between owners.
Power cost and revenue split are conditional averages
The reported US$0.043/kWh average all-in power cost is defined as a weighted figure assuming all installed computing power is energized, excluding JV calculations. The 60.5% average revenue split similarly describes the weighted share Canaan would receive under applicable joint-mining arrangements at full energization, excluding JV ownership. Neither is a promise that every site pays one tariff or that 60.5% of an independently calculated gross return reaches the balance sheet.
For a farm comparison, the practical distinction is between a modelled weighted rate and an actual bill tied to a meter and interval. Installed MW multiplied by every hour of the month would assume operation that the release’s definitions do not establish. The company’s average also cannot replace another customer’s hosting price in our calculator. A comparable economic result would need actual work, energy, allocation, charges and the same period for all inputs.
Fleet efficiency does not identify a new ASIC configuration
Global average miner efficiency is reported as 23.7 J/TH for non-JV equipment and 24.6 J/TH for JV equipment. Installed power capacity is shown separately as 244.9 MW non-JV and 120 MW JV. These are fleet and infrastructure aggregates. They do not identify one new machine model, its factory bin or a measured power profile that can be copied into the hardware catalogue.
As a simple period calculation, 44 BTC over September’s 30 days averages approximately 1.47 BTC per calendar day. That is an editorial average of reported production, not evidence of constant daily output. It also does not remove the JV exclusion or paused-site qualification. A device profitability estimate still needs its own hashrate, power, operating conditions and network assumptions rather than a company-wide average with a different ownership boundary.
Heat-reuse milestones need their own units
Canaan’s confirmed operational statement concerns the Nordic district-heating project, described as 8 MW and serving about 2,800 homes with recovered computing heat. Those numbers do not provide a measured heat-delivery curve for every home or a season-long utilization factor. The release does not establish that the entire named capacity is useful heat at a customer meter in every hour.
For a heat-reuse comparison, useful evidence would include computing input, recoverable thermal output, delivered heat, temperature and the demand period. Counting the same energy once as electrical input and again as newly generated electricity would be incorrect. The Canadian greenhouse project should remain at the inspection-dependent planning stage until evidence supports deployment. Nordic operation does not, by itself, complete the separate Canadian project or validate all future heat-sales assumptions.
Treasury monetization has a later cut-off
As of October 15, Canaan says it had monetized 3,952 ETH and 54.5 BTC for approximately $13.9 million. That combined cash total does not disclose a Bitcoin-only sale price. The release separately mentions $5.4 million used to repurchase 13.6 million ADSs in late August and approximately 16.4 million ADSs repurchased year to date. The different periods and transaction scopes should not be collapsed into a claim that the full $13.9 million has already been spent on repurchases.
Month-end Bitcoin holdings include receivable coins and exclude customer deposits under Canaan’s definition. A balance-sheet amount is therefore not simply a count of coins available in one public wallet. September production, later monetization and share repurchases cannot be reconciled by subtracting a single pair of headline numbers. The release supports the stated transactions and dates, not a reconstructed complete cash-flow or custody ledger.
What the next comparison should preserve
The important development is an updated mining report alongside a now-operational Nordic heat project, with Canada still conditional. Future comparisons should keep the reporting period, ownership, energized-capacity definition and actual accepted work visible. Those distinctions are essential when checking a changing fleet, rather than treating every number carrying EH/s as interchangeable.
The photographs are archival mining and district-heating illustrations, not proof of the company’s current sites or the new project’s measured performance. Our calculations are labelled averages or snapshot ratios. No new ASIC specification, guaranteed site tariff, constant production rate or investment return is inferred. Existing calculator inputs and catalogue records remain separate from this corporate operating update.
Source: Canaan ↗
Mining calculator ↗
