Public bitcoin miners lost 21% of hashrate: the bet on AI is reshaping the industry
The public Bitcoin mining sector is undergoing a structural transformation. Over the past three quarters, the realized hashrate of leading companies (excluding Bitdeer) has declined by 21.2% — from 368.3 EH/s in Q4 2025 to 319.0 EH/s in Q2 2026. This is not merely a correction, but a deliberate dismantling of capacity to pivot toward high-margin segments.
The network is declining more gently
The overall Bitcoin network shows a more moderate decline: the average quarterly hashrate fell from 1,071 EH/s to 957 EH/s, representing only 10.6%. A key factor smoothing the drop is Bitdeer's aggressive growth, whose realized hashrate surged by 44% to 63 EH/s. Notably, some companies are winding down mining faster than others can scale up their volumes, creating a temporary imbalance.
The main driver is economics. Revenue from colocation and computing infrastructure for AI is becoming significantly more attractive than traditional mining. Core Scientific earned $136.7 million in Q2 from additional colocation deals versus a meager $27.5 million from Bitcoin mining. TeraWulf shows an even more radical skew: $31.9 million from HPC leasing (71% of revenue) versus $12.8 million from mining.
Uneven transition
However, the picture is not uniform. Riot Platforms is only partially diversifying so far — $23.2 million from data centers versus $113.7 million from mining. Bitdeer earns $14 million from cloud AI services, but its core revenue ($197.1 million) still comes from mining. Hut 8 and MARA modestly note contributions from computing services, while Cipher and Keel Infrastructure have not yet even begun recording HPC revenue.
The current downturn is a classic effect of weak mining economics and intense competition for capital and electricity from AI workloads. For comparison: after China's mining ban in June 2021, the network hashrate briefly collapsed to 57.5 EH/s but recovered by December. The current cycle is different — it is not an external shock, but an internal evolution of business models.
My analysis: the market is witnessing not capitulation, but a fundamental shift. Miners that first convert their energy contracts and infrastructure into AI capacity will gain a strategic advantage. Those that remain purely Bitcoin-focused will face margin pressure. The question is not whether mining will survive, but who will become its dominant player in the era of hybrid data centers. Given that fee revenue in August 2026 fell to a decade low, the bet on diversification looks not just justified, but inevitable.