35% of Canaan’s July hashrate listed from paused Ethiopia

Canaan reported 14.24 exahashes per second of operating computing power for July. The company’s table assigns 4.96 EH/s to two projects in Ethiopia, representing roughly 35% of the reported operating total, while footnote 10 states that mining in Ethiopia had been paused.

The July table lists 4.96 EH/s for Ethiopia in both the operating and installed columns. The filing does not state whether the paused machines were energized or producing hashes as of July 31, leaving the level of active capacity in Ethiopia unspecified.

Canaan defines operating computing power as the theoretical output of energized mining machines assuming all were operating. That definition can include machines that are temporarily offline and applies only to the company’s non-joint-venture operations.

Subtracting the 4.96 EH/s attributed to Ethiopia from the rounded global total produces 9.28 EH/s. The company’s table classifications mean that subtraction is an exclusion calculation based on reported categories rather than an adjusted live-hashrate figure.

The filings show a change from the end of June. On June 30 Ethiopia appeared with 0.36 EH/s listed as operating against 4.96 EH/s installed; the filing attributed that June drop primarily to one site going temporarily offline in mid-June for local power-grid maintenance. In the July table, the full 4.96 EH/s returned to the operating column while footnote 10 described Ethiopian operations as paused. The documents do not indicate whether the July pause had the same cause or whether it was expected to be temporary.

Canaan reported mining 46 BTC in July and held 1,917 BTC and 3,952 ETH at month-end. The company separately reports metrics for its 49%-owned Alborz, Bear and Chief Mountain joint ventures in West Texas. Joint-venture output is excluded from the company’s Bitcoin production and average all-in power-cost calculations even though joint-venture capacity appears in the global projects table.

The filings do not specify whether Ethiopian machines were actively hashing at month-end, so their contribution to July’s production cannot be determined from the reported tables alone.