The Bitcoin mining sector is evolving. Together, the surge in costs, the collapse of fees, and the rise of artificial intelligence are forcing miners to rethink their strategy. What once transformed stable operations has thus turned into a true battleground for the computing power of the next generation.
The harsh reality of Bitcoin mining
Coinshare's report on Bitcoin mining for the fourth quarter of 2025 revealed that the sector has reached a true breaking point. Production costs have skyrocketed to unprecedented levels, the hash price has collapsed, and artificial intelligence (AI) is now outbidding miners for their own infrastructure, resulting in the largest structural upheaval ever seen in the sector.
The crypto mining industry entered the second quarter of 2025 with a ruthless new reality:
The average cash cost to mine one Bitcoin for public miners has risen to about $74,600,
total costs have risen to $137,800,
transaction fees, which previously served as a buffer for miners' revenues, fell to less than 1% of block rewards in May and June, the lowest contribution since the 2024 halving.
And yet, even as margins collapsed, the Bitcoin network continued to make progress, surpassing for the first time the 1 Zetta hash/s mark in August.
Public miners have only contributed about 80 EH/s to the cumulative growth for the year, with most of the expansion now coming from private operators, sovereign miners, and well-capitalized energy players with much lower energy costs.
Result: miners are experiencing dilution due to hash rate growth that they no longer control.
AI is taking hold and generating 10-20 times more per megawatt
A much larger disruption is occurring at the infrastructure level. Industrial-scale mining campuses, which include sites from 100MW to 1GW, share almost the same requirements for power, cooling, and rack density as modern AI-enabled data centers.
This overlap has made mining facilities prime targets for hyperscalers.
Agreements between Google and TeraWulf, Google and Cipher, and multi-site agreements with Fluidstack all point in the same direction: the large-scale tech sector is investing heavily in capabilities initially built by miners.
The numbers explain the reason. Bitcoin mining generates about $1 million per megawatt, while AI-based computing generates between $10 and $20 million per megawatt. No miner can ignore this margin.
The division of the industry: AI megacampus vs. very low-cost mobile miners
The sector is now divided into two clear models:
1. Mega-scale miners → fully or partially convert to AI/HPC
These centers can upgrade their electrical topology and availability standards to meet enterprise requirements. They sign contracts for several decades and transition from volatile block rewards to stable revenues based on capacity.
2. Low-cost, mobile miners → turn to isolated energy
Miners unable to compete with AI are moving off-grid: flare gas, remote hydro, and surplus renewable energies. Portable platforms are being deployed wherever cheap energy exists, evoking the decentralized beginnings of mining.
This migration marks a long-term transformation of the industry, not a temporary cycle.
According to the CoinShares report:
The hash price averaged around $50 per PH/s/day throughout the second quarter, continuing its post-halving decline.
With rising difficulty, stagnant fees, and Bitcoin trading primarily within a margin, the oldest groups of ASICs have been taken offline.
Analysts expect hash price to remain in a range of $37 to $55 per PH/s/day until 2028, unless Bitcoin experiences a much faster rally than hash rate growth.
A structural shift: AI surpasses Bitcoin
For the first time in Bitcoin's history, miners find themselves excluded from their own infrastructure.
Indeed, the economic superiority of AI, the flow of agreements from hyperscalers, and the rising costs of industrial mining are driving the sector towards a permanent transformation.
The Bitcoin network, however, remains robust, with hash rate still on the rise, but the mining sector is transforming rapidly.
This places miners at a key turning point: either invest massively in AI or migrate to isolated energy sources.
The moral of the story: the end of the world begins when the robot renders the human obsolete.
