Bitcoin network regains 1 ZH/s as mining capacity shifts beyond public operators
Bitcoin’s seven-day average hashrate has climbed back above 1 zettahash per second even as major listed miners keep reallocating capacity toward AI and high-performance computing. The rebound suggests replacement supply is emerging from expanding operators such as Bitdeer and from smaller buyers gaining access to both new and used mining equipment.
Highlights
- Bitcoin's seven-day average hashrate surpassed 1,000 EH/s for the first time since March 3, despite public miners reallocating capacity to AI.
- Bitcoin mining difficulty rose 5.92% on October 16, with another estimated 5.16% increase expected in about seven days, signaling greater network competition.
- Bitdeer expanded self-mining and co-mining capacity to a combined 111 EH/s in September, while resale channels enable smaller operators to absorb equipment from larger firms exiting mining.
Hashrate recovery offsets public miners’ AI shift
TheEnergyMag reports that Bitcoin’s seven-day average hashrate has moved back above 1,000 EH/s over the past week, reaching its highest level since March 3. The increase stands out because public mining companies have been reducing realized hashrate while redirecting infrastructure and power toward AI workloads.That recovery is also feeding into network difficulty. Bitcoin mining difficulty rose 5.92% on October 16, and block production in the current difficulty epoch points to another estimated 5.16% increase in about seven days, although that figure can still change as more blocks are added.
The shift away from mining is not easily reversible once sites are converted. When machines are removed and facilities are rebuilt for AI customers, that power capacity is no longer positioned to return quickly to Bitcoin mining if economics improve.
Against that backdrop, the latest hashrate rebound indicates the network is replacing lost public-miner capacity from other sources rather than simply benefiting from a temporary pause in AI conversion plans.
Smaller operators and hardware resale support competition
One visible contributor is Bitdeer, whose September update showed self-mining capacity at 82.3 EH/s, up from 79.9 EH/s in August and 35 EH/s a year earlier. Its co-mining business, which places Bitdeer-owned machines in third-party data centers, also expanded to 28.7 EH/s from 21.6 EH/s, adding 9.5 EH/s across the two categories in a month even as the company also pursues AI opportunities.The article also points to a quieter source of recovery, smaller and mid-sized operators acquiring equipment that larger miners no longer need. Machines removed from sites leaving Bitcoin mining can return to service if they are relocated to buyers with lower electricity costs, available rack space, and a more attractive purchase price.
Resale channels appear to support that trend. Sazmining, for example, lists pre-owned Antminer S21 XP machines for hosting in Paraguay and Texas, illustrating how used hardware can re-enter the market even if the original owners are not identified.
The same contrast is emerging in newer equipment. Core Scientific terminated its equipment agreement with Block and Proto Global in the second quarter and recorded a $41.9 million loss as it shifted toward high-density colocation, while the original agreement covered chips representing 15 EH/s. At the same time, Sazmining is offering individual buyers access to Proto immersion hashboards deployed at a U.S. site, reinforcing the view that hardware once concentrated among large miners is increasingly reaching retail and smaller-scale participants.
With one sharp difficulty increase already recorded and another taking shape, the renewed flow of machines and capacity back into the network is restoring competitive pressure across the mining sector.
Our earlier article on the U.S. AI data center boom explained that the rapid expansion is running into rising financing costs, construction delays, and tightening power constraints. We highlighted that a sizable share of planned projects may prove unrealistic as investor sentiment weakens and developers face local opposition and grid limitations—factors that can slow how quickly AI compute capacity actually comes online.