Yes, a portion of miners is finding it more convenient to direct infrastructure and energy toward AI data centers. And yes, Bitcoin's hashrate is down about 20% from its highs. But saying Bitcoin's security is in danger is a stretch: mining works through a system of incentives and rebalancing that has always brought the network to new equilibria.
The data
According to BlocksBridge Consulting, realized hashrate among a cohort of public miners fell from 368.3 exahashes per second (EH/s) in Q4 2025 to 319 EH/s in Q2 2026, a 13.4% decline. Excluding Bitdeer, which kept expanding (+44%, to 63 EH/s), the drop reaches 21.2%: from 324.6 to 255.9 EH/s.
Over the same period, the Bitcoin network's average hashrate fell 10.6%, while difficulty dropped 19.9% from the peak of about 156 trillion reached in November 2025: according to cryptonews.net, it is one of the three deepest declines since the ASIC era.
The clearest signal comes from balance sheets. Core Scientific generated $136.7 million in colocation revenue in Q2, against $27.5 million from Bitcoin mining. TeraWulf booked $31.9 million from HPC lease revenue, versus $12.8 million from mining.
Why miners are choosing AI
AI data center contracts pay more than mining, and more predictably. With production costs estimated near $90,000 per bitcoin and a spot price around $67,000, mining margins are negative: many listed operators have therefore redirected energy and sites toward AI and high-performance computing, funding the transition with debt and bitcoin sales.
This is not the first time hashrate has collapsed. After China's 2021 ban, the network lost a huge share of computing power only to recover quickly, as miners relocated abroad.
Why mining is not in danger
Mining works through incentives and rebalancing. When mining is highly profitable, new operators enter, competition increases and difficulty rises. When margins compress, some exit, hashrate falls and difficulty later adjusts, until a new equilibrium is reached.
The figure that puts everything in perspective: the last time Bitcoin traded at similar price levels, hashrate was about half of what it is today. The network is much stronger than the price alone suggests.
There is also a decentralization argument: US-listed miners have accounted for over 40% of global hashrate, and a reduction in their weight could make the network more geographically distributed. CoinShares, despite the contraction, still forecasts hashrate growth to about 1.8 zettahashes per second by end-2026, conditional on bitcoin recovering toward $100,000.
Contractions are part of the system. What really matters is the long-term trend: Bitcoin has survived every cycle, and mining will keep existing as long as there are incentives to protect the network.
Sources