Bitcoin mining difficulty reached a fresh all-time high in 2026, while network hashrate moved past 1 ZH/s. The network kept expanding, but miner economics stayed under pressure. After the April 2024 halving cut block rewards in half, revenue per unit of computing power weakened sharply; by the time Bitcoin had fallen from its October 2025 peak, hashprice had dropped 66% from that high.
Older machines lost viability after the halving
According to the source material, miners running hardware above roughly 25 J/TH were operating at a loss in most power markets after the halving. The efficiency shift has been steep: over eight years, mining hardware improved from about 98 J/TH to below 15 J/TH, a roughly 7x gain in energy efficiency. One example cited is the Antminer S21 XP at 13.5 J/TH, which can deliver the same output as several older rigs using the same power.
That gap has reset the economics of survival. Large pools such as Foundry USA and F2Pool still control a major share of global hashrate, while smaller operators are joining pools or shifting to cloud-based setups to remain viable. MARA and CleanSpark responded by reinvesting in infrastructure and holding onto the bitcoin they mined instead of selling it.
Production costs climbed while daily revenue sank
JPMorgan estimated the industry’s average production cost at about $77,000 per BTC. The most efficient miners, using electricity below $0.05/kWh and the newest ASICs, can reportedly produce bitcoin for roughly $34,000 to $43,000 per BTC. For the rest of the market, the economics are far tighter. In late January 2026, daily miner revenue fell to about $28 million, and the profitability index touched a 14-month low.
Price remains the variable that matters most. When Bitcoin reached $122,000 in July 2025, miners were earning more than $380,000 per block even after the halving. The source also lists a spot price of $78,203.48 and a market capitalization of roughly $1.56 trillion. The same network can look very different depending on where BTC is trading.
The difficulty spike is being read as a signal from surviving miners
Mining has not stopped because difficulty adjustment still offers a path to recovery for those left standing. When weaker miners exit, difficulty can ease and margins can improve for more efficient operators. The source cites Phemex, which said the February 2026 difficulty jump to 144.4T suggests surviving miners have rebuilt capacity and pushed the network into Phase 4 of a historical five-stage cycle seen when BTC trades below production cost.
In that reading, price recovery has tended to come only after miner capitulation runs its course. What stands out here is not a forecast but the tension in the data: revenue has weakened, yet network strength has not broken. Difficulty and hashrate are still elevated, which points to resilience among operators with the lowest costs and strongest hardware fleets.
Mining companies are also redirecting infrastructure toward AI
Some listed miners are now using data center assets for AI-related business lines. Bitfarms has removed “bitcoin” from its corporate name, and Riot Platforms is facing activist pressure to expand AI operations. Phemex argues that the logic is simple: in some markets, AI hosting generates more revenue per megawatt than bitcoin mining.
That shift is accelerating consolidation across the sector. Industrial-scale operators with cheap electricity, next-generation ASICs, and deeper infrastructure budgets are widening their lead. Smaller and less efficient miners are left to merge into pools, move to alternative hosting models, or exit. The next difficulty adjustment will show whether this record level can hold or whether another reset is still ahead.