Bitcoin difficulty could drop 16% as AI leases pull miners

An automatic Bitcoin difficulty adjustment scheduled for about July 26 could reduce difficulty by roughly 16%, increasing the expected share of block rewards for machines that remain online. The network lowered difficulty 5% on July 11 to 127.17 trillion.

Market data showed hashprice — the daily revenue per PH/s — at $30.88 on July 13 with a seven-day average of $30.39. That level is at or below breakeven for many operators depending on power costs and machine efficiency. Hashprice recovered from roughly $27.60 in early July but remained about 37% below an October 2025 peak near $49.40. Transaction fees provided minimal additional revenue, accounting for about 0.69% of block rewards in the week to July 13.

Bitcoin’s protocol adjusts difficulty every 2,016 blocks to keep the average block time close to ten minutes. When total computing power falls, the protocol lowers difficulty so each unit of hashrate receives a larger expected share of the 3.125 BTC subsidy per block. The protocol’s correction affects block times and expected rewards but does not change miners’ contracted electricity rates, debt obligations, or the resale value of older equipment.

Difficulty has been volatile in 2026. Compounded difficulty fell about 14.22% from a January 8 peak of 146.47 trillion. Eight of the first 14 retargets this year were negative. The series of adjustments included a 10.09% drop in mid-June, a 7.15% rise on June 26, and a 5% fall on July 11, reflecting machines leaving and returning to the network.

A double-digit reduction would improve unit economics for online machines, but the benefit would not be uniform. Operators with newer, more efficient ASICs, lower power costs and access to capital are best positioned to capture the upside. Operators with long-term, high-cost power contracts, heavy debt or constrained liquidity may remain unprofitable after the retarget.

Public company disclosures illustrate the split. CleanSpark reported about 50 EH/s operational in June, produced 614 BTC that month and sold 429 BTC; the company listed 13,924 BTC on its balance sheet while recording 1,719 BTC as collateral or receivables related to derivatives. Marathon sold 20,880 BTC for roughly $1.5 billion in the first quarter of 2026, reported a $1.26 billion net loss on $175 million of revenue, reduced staff by about 15% and recorded $45.9 million in restructuring charges. Riot Platforms reported 42.5 EH/s deployed and produced 1,473 BTC in the first quarter while selling 3,778 BTC for net proceeds of $289.5 million and establishing contracted compute capacity at its Rockdale site.

Several miners have signed large AI and high-performance compute leases. TeraWulf entered a 20-year lease with an AI tenant covering about 401 MW and expected contract revenue near $19 billion. Hut 8 expanded a Beacon Point lease to raise campus capacity to 704 MW and a base-term contract value near $19.6 billion. These deals provide long-term, predictable payments for power and infrastructure at those sites.

Research on the Texas electricity market found that mining load declines as power costs rise, but the degree of curtailment depends on hashprice: higher expected mining revenue makes operations more willing to run at higher wholesale prices. Operators that can temporarily ramp down provide some flexibility during price spikes, but decisions hinge on the relationship between electricity price and the revenue available from hashing at the time.

What follows the retarget will be observed in short-term signals. Analysts and operators will watch post-adjustment hashprice, production updates from public miners, miner-to-exchange transfers, shifts in mining-pool share and the estimated direction of the next epoch. The difficulty mechanism will continue to adjust block timing, while power contracts, balance sheets and signed AI compute leases will shape how electricity and infrastructure are allocated across uses.