As AI reshapes mining, the real reward for Bitcoin miners is power
Bitcoin miners spent years racing to secure cheap electricity, which has since become more valuable than the Bitcoin mining business it was built on.
This reversal drives Fidelity’s May 2026 assessment that AI hosting could level Bitcoin’s hash rate as major operators reorient their energy infrastructure away from pure mining while providing a second revenue stream for miners, with two hyperscaler contracts putting specific prices on what miners build.
Cipher Mining’s SEC filing business update announced a roughly $5.5 billion, 15-year lease agreement with AWS to provide 300 MW of turnkey space and power for AI workloads, with delivery beginning in July 2026.
IREN signed a five-year GPU cloud contract with Microsoft worth approximately $9.7 billion to deploy NVIDIA GB300 GPUs at its 750 MW Childress, Texas campus through 2026 and support 200 MW of critical IT load.
| pitman | hyperscaler | contract value | continue | Power/Capacity | Delivery Schedule | Why is it important? |
|---|---|---|---|---|---|---|
| password mining | AWS | ~$5.5 billion | 15 years | 300MW | Starting in July 2026 | Demonstrates that power mining sites can be leased with AI infrastructure. |
| irene | microsoft | ~$9.7 billion | 5 years | 200 MW core IT load on 750 MW Childress campus | GPUs Deployed by 2026 | It shows that miners can monetize Power Campus not only through BTC mining but also through GPU Cloud. |
Miners have already secured the land, grid interconnection, substations, and power rights needed for AI data centers, but they can’t build them fast enough.
Compressed hash prices in 2024 increased CoinShares’ tracked weighted average cash cost to approximately $79,995 per BTC by Q1 2026. This has led operators to push AI hosting as a revenue stabilizer, leasing unused capacity, keeping mining rigs running, and offsetting the worst of the Bitcoin downturn.
CoinShares estimates that public miners’ AI and HPC contracts will total more than $70 billion by early 2026, with listed miners on pace to generate between about 30% and up to 70% of their revenue from AI by the end of the year.
This is a revenue hedge since the Cipher and IREN contracts were replaced by price discovery for the power campus.
Price discovery changes internal math
A January 2026 analysis by Fidelity found a mining and AI crossover of around $60 to $70 per petahash per day on a fleet of 20 lines per terahash. This means that hash prices would need to increase by 40-60% for most 20-25 J/TH miners to match the contracted GPU hosting economics.
Hashrate Index’s May 25 data has since extended this distance to US dollar-denominated hash prices at $35.88 per PH/day, placing the AI crossover approximately 67% to 95% above its current point.
A miner sitting on 300 MW of powered and permitted infrastructure now has to choose between deploying an ASIC and earning $35.88 per PH/day or signing a hyperscaler lease with a contract fee that should almost double the hash price.
AWS and Microsoft have effectively announced a floor for what their infrastructure is worth to anyone other than Bitcoin, and every major operator with similar assets now has that figure in their models.
Building AI infrastructure costs between $8 million and $15 million per megawatt, compared to $700,000 to $1 million for Bitcoin mining infrastructure, and converting miners will be entering a more capital-intensive business with a fundamentally different debt profile, valuation metrics, and execution risk.
Hash rate may no longer follow BTC price alone.
Bitcoin’s mining expansion has historically followed price, with miners ordering more machines when BTC rises and reducing capacity when it falls.
VanEck’s April ChainCheck recorded 30-day hash rate momentum at the 16th percentile and 90-day momentum at the 9th percentile. This is the densest cluster of sustained hash rate declines following China’s 2021 mining ban.
According to CoinWarz data as of May 28, Bitcoin difficulty was 136.61T and the 90-day difficulty change was -5.40%. This is consistent with Fidelity’s mining exodus picture.
Bitcoin’s 2,016 block difficulty adjustment is still a counterweight because it lowers the computational cost of generating a valid block each time the hash rate expires and increases the return per hash unit remaining once the difficulty resets.
If the hash rate exits at 20%, the hash price for surviving miners rises to roughly $44.85 per PH/day, and if it exits at 30%, it rises to around $51.26, which is still well below Fidelity’s AI crossover unless the BTC price or transaction fees rise meaningfully.
Power locked into a 15-year AWS lease or 5-year Microsoft GPU contract cannot return to mining even when ASIC economics recover. In previous cycles, idle hashes were returned because the computer could be turned on again, but in this cycle, the campus itself may be committed elsewhere.
Bitcoin further narrows the market needed.
If BTC moves from $100,000 to $140,000 or transaction fees rise significantly, the economy will be reshaped.
A 20% decrease in network hash rate would lower the BTC price needed to hit the $60-$70 AI crossover to around $98,000-$114,000, while a 30% decrease would lower that threshold to around $86,000-$100,000.
Miners still committed to Bitcoin benefit from a market where hash prices rise faster than hash rates, compressing the playing field and improving margins for operators with efficient equipment and lower power costs.
Fewer large public miners in the hashrate mix will also mean fewer forced BTC sales, which have historically put pressure on spot prices during expansion cycles.
Charles Schwab’s May 26 analysis argues that the hybrid infrastructure model strengthens Bitcoin’s overall network health. That is, lower forced sales, stricter difficulty conditions, and better miner margins reduce the systemic stress that large, capital-intensive miners have historically introduced at cycle peaks.
The industry is divided into two distinct businesses: companies that own powerful campuses and monetize them through hyperscaler contracts, and companies that actually mine Bitcoin at lower-cost, more flexible or isolated energy sites that AI data centers cannot easily operate.
| script | Hash Rate End | Implied hash price after difficulty reset | BTC price required for $60/PH/day | BTC price required at $70/PH/day | takeout |
|---|---|---|---|---|---|
| maintain the status quo | 0% | $35.88 | ~$122,000 | ~$142,000 | Mining is much lower than AI crossover. |
| Normal end | 20% | ~$44.85 | ~$98,000 | ~$114,000 | Difficulty resets help miners, but they don’t completely close the gap. |
| bigger exit | 30% | ~$51.26 | ~$86,000 | ~$100,000 | As BTC rises or fees improve, Bitcoin mining becomes more competitive. |
AI wins allocation decisions
As BTC remains below $70,000-$80,000, fees remain low, and power prices remain high, contracted GPU hosting economics will dominate the internal capital allocation of operators with AI-enabled sites.
CoinShares estimates that for machines with S19 XP efficiency or less, power costs above $0.06 per kilowatt-hour would be around $30 per PH/day, making 15% to 20% of the world’s fleet uneconomical.
Older fleets are shut down, difficulty decreases over successive eras, and surviving miners earn more per petahash. However, it is still not enough to close the gap with the Cipher and IREN contracts for operators who still have options.
Difficulty adjustments keep the network running through all exits, and the center of gravity of mining shifts as large public miners with AI-enabled infrastructure become data center tenants, while Bitcoin hash rates are concentrated on operators with cheaper, intermittent or internationally diverse energy.
The IREN/Microsoft contract includes an explicit delivery schedule clause, reported by Reuters, that could trigger termination if a milestone is missed, and miners saddled with massive debt along with delayed AI revenues will face asset price adjustments from Bitcoin proxies to run-risk assets.
Division is the result
The competition between ASICs and GPUs for miner capital will be site-by-site and operator-by-operator, depending on already concluded power contracts and the BTC price at the next halving.
Bitcoin’s network absorbs hashrate exits through lower difficulty, while higher BTC prices or fees can drive economics out of mining for all operators who haven’t yet authorized it elsewhere.
A more sustainable outcome of the AWS and Microsoft deal is that it will now be possible to run large-scale, reliable and profitable infrastructure businesses on the same sites that Bitcoin miners built, without having to mine a single block.
Whether that possibility becomes the default for building the next power campus will depend on where the BTC price settles relative to $35.88 and how many hyperscalers arrive with 15-year checkbooks before the next halving puts the question back in question.
