Bitcoin's network hashrate hovers at 1.004 ZH/s (zettahash per second) as of April 2026, a symbolic milestone that masks a structural fracture in mining economics. The network's computational power crossed 1 ZH/s for the first time in December 2025, but the number obscures a more consequential tr...
"We expect further capitulation among higher-cost operators in H1 2026 unless BTC's price recovers materially." — CoinShares, Q1 2026 Bitcoin Mining Report
Bitcoin's network hashrate hovers at 1.004 ZH/s (zettahash per second) as of April 2026, a symbolic milestone that masks a structural fracture in mining economics. The network's computational power crossed 1 ZH/s for the first time in December 2025, but the number obscures a more consequential trend: publicly listed miners sold over 32,000 BTC in Q1 2026 — more than in all of 2025 combined — while signing over $70 billion in cumulative AI and high-performance computing (HPC) contracts that redirect electrical capacity away from Bitcoin.
Hash price, the standard measure of mining revenue per unit of computational power, fell to $29/PH/s/day in Q1 2026, a five-year low. The weighted average cash cost to produce one bitcoin among public miners reached approximately $79,995 in Q4 2025, according to CoinShares data. With Bitcoin trading near $78,000 in late April, margins for the median miner are effectively zero or negative. The result is an accelerating pivot: listed miners could derive as much as 70% of their revenue from AI infrastructure by year-end, up from roughly 30% today, according to CoinShares' Q1 2026 report.
This dynamic creates a measurable tension in Bitcoin's security model. If AI workloads generate 3x to 25x more revenue per megawatt-hour than Bitcoin mining, rational capital allocation dictates reallocation. Bitcoin's hashrate posted its first quarterly decline in six years during Q1 2026, falling approximately 4% year-to-date. The question is no longer whether miners will pivot to AI — they already have — but whether Bitcoin's security budget can absorb the loss.
Bitcoin's hashrate — the aggregate computational power securing the network — reached 1 ZH/s in December 2025 and hit multiple all-time highs above that threshold in January 2026, according to CoinWarz data. The metric represents 1,000 exahash per second, or roughly 1 sextillion hash computations every second.
The milestone was short-lived. Severe U.S. winter storms in January 2026 forced Texas-based miners to curtail operations, triggering a 12% hashrate drawdown — the largest since China's 2021 mining ban, according to CryptoQuant. Daily mining revenue plunged from $45 million to $28 million in two days. Output from large public miners dropped from 77 BTC per day to 28 BTC per day.
By late March, hashrate recovered to approximately 1.02 ZH/s, according to CoinReporter. As of April 18, BT-Miners reported the network at 995 EH/s. The current reading on CoinWarz shows 1.004 ZH/s.
Mining difficulty — the network's self-adjusting mechanism that calibrates block production — experienced three consecutive negative adjustments in Q1, the first such streak since July 2022, signaling miner capitulation. Difficulty fell 2.43% to 135.59 trillion on April 17, according to CryptoNews. The next adjustment is projected for approximately April 30, and average block intervals of 9 minutes 35 seconds suggest difficulty will increase.
The post-halving compression continues to tighten. Bitcoin's fourth halving on April 20, 2024, cut the block subsidy from 6.25 BTC to 3.125 BTC. Transaction fees have failed to compensate, running consistently below 1% of total block rewards throughout Q1 2026.
Key metrics from CoinShares' Q1 2026 Bitcoin Mining Report:
| Metric | Value | |---|---| | Hash price (Q1 2026) | ~$29/PH/s/day (five-year low) | | Breakeven hash price (industry average) | ~$35/PH/s/day | | Miners in unprofitable territory | ~15-20% of legacy fleet | | Weighted avg. cash cost per BTC (Q4 2025) | ~$79,995 | | Fully loaded cost per BTC | ~$137,800 | | BTC spot price (April 22, 2026) | ~$78,300 | | Public miner BTC sales (Q1 2026) | 32,000+ BTC |
The $35/PH/s breakeven threshold means approximately 20% of the global mining fleet is operating at a cash-flow loss at current hash prices. Efficient operators running Antminer S21 XP hardware can remain solvent down to approximately $55,000 BTC, according to industry estimates. Older fleets begin shutting down below $75,000.
MARA Holdings sold 13,210 BTC in Q1. Riot Platforms divested 4,026 BTC. Core Scientific liquidated approximately 1,992 BTC ($175 million) in January alone, according to CoinDesk. Publicly listed miners have collectively reduced BTC treasuries by over 15,000 BTC from peak levels.
The economic logic is straightforward. One megawatt allocated to Bitcoin mining generates volatile, shrinking revenue. The same megawatt leased to a hyperscaler under a multi-year AI infrastructure contract delivers predictable, fixed-rate income at 3x to 25x the revenue per megawatt-hour, according to S&P Global research.
Over 2025 and early 2026, publicly listed miners signed more than $70 billion in aggregate GPU co-location and cloud service contracts. The largest commitments:
Core Scientific: A 12-year, $10.2 billion agreement with CoreWeave for approximately 590 MW of HPC capacity across multiple sites. The company is raising $3.3 billion via a junk bond sale, announced April 21, to finance conversion of its 1.2 GW total capacity toward AI data centers. Core Scientific is selling the majority of its Bitcoin holdings by end of 2026 to fund the transition.
IREN: A $9.7 billion contract with Microsoft for 76,000 NVIDIA GB300 GPUs across 200 MW at its Childress, Texas campus.
TeraWulf: HPC contracts valued at $12.8 billion, with 27% of recent quarterly revenue already derived from non-mining operations.
Hut 8: A 15-year, $7 billion lease with Fluidstack (backed by Google) for initial deployment of 245 MW at its River Bend campus in Louisiana. The partnership extends to Anthropic.
Riot Platforms: Signed a leasing agreement with AMD and acquired land at its Rockdale, Texas site, marking its first hyperscale data center tenant.
The debt required to fund these transitions is substantial. IREN carries $3.7 billion in convertible notes. TeraWulf's total debt has reached $5.7 billion. CIFR issued $1.7 billion in senior secured notes. These are not mining companies that also do AI — they are rapidly becoming infrastructure companies that still happen to mine Bitcoin.
Bitcoin's security model rests on an economic assumption: miners receive sufficient compensation — block subsidies plus transaction fees — to make honest mining more profitable than attacking the network. This compensation is commonly referred to as the "security budget."
Three converging trends threaten this assumption in 2026:
1. Revenue plateau. Post-halving block subsidies are at 3.125 BTC ($244,000 at current prices) per block, down from 6.25 BTC pre-April 2024. Transaction fees have not scaled to compensate.
2. Opportunity cost escalation. The $70 billion in AI contracts creates a measurable alternative for every megawatt currently dedicated to Bitcoin. When AI revenue per MW exceeds Bitcoin revenue per MW by an order of magnitude, the rational allocation is clear.
3. Hashrate concentration. The top ten mining operations now control approximately 45-50% of total network hashrate, up from 35% eighteen months ago, according to industry data. The top three countries (U.S., China, Russia) control roughly 68% of global hashrate. Geographic and corporate concentration increases vulnerability to coordinated disruption, whether through regulation, weather events, or infrastructure failures.
Bitcoin's hashrate posted its first quarterly decline in six years during Q1 2026, dropping approximately 4% year-to-date, according to CoinDesk. CoinShares forecasts the network will reach 1.8 ZH/s by end of 2026 and 2 ZH/s by end of March 2027 — but that projection assumes Bitcoin recovers to $100,000 by year-end. At current prices, that forecast is unlikely to materialize.
The efficiency gap between best-in-class miners (~15 watts per terahash) and lagging fleets (~25+ W/T) is wide enough that acquiring efficient capacity may be cheaper than upgrading legacy equipment. Further M&A activity is expected through H2 2026.
Marathon Digital Holdings leads public miners with 30.6 EH/s of installed hashrate and 50,639 BTC in reserves (as of mid-2025). CleanSpark, Riot Platforms, and Core Scientific follow. Emerging-market entrants — Paraguay, Ethiopia, Oman — have entered the global top 10 mining jurisdictions, driven by operators like HIVE (300 MW in Paraguay) and BTDR (40 MW in Ethiopia), according to Hashrate Index data.
The geographic diversification is marginal relative to the concentration risk. If U.S.-based miners continue redirecting capacity to AI, the resulting hashrate reduction will not be fully offset by emerging-market additions, which lack comparable electrical infrastructure and capital access.
Bitcoin's 1 ZH/s hashrate is a measurement, not an achievement. The network's raw computational power has never been higher, but the economic model underwriting that power has never been more strained. Miners are rational actors. When AI infrastructure contracts deliver multi-year, predictable returns at multiples of Bitcoin mining revenue, capital flows accordingly.
The security budget question is no longer theoretical. Three consecutive negative difficulty adjustments in Q1, the first such sequence in nearly four years, signal that hashrate is leaving the network. Whether this constitutes a cyclical adjustment or a structural shift depends on Bitcoin's price trajectory. At $78,000, the math does not favor continued mining expansion. At $100,000, CoinShares models suggest hashrate resumes growth. The spread between those two numbers — roughly 28% — represents the gap between Bitcoin's current security budget and the one it needs.
The mining industry is not dying. It is transforming into something different: a hybrid infrastructure sector where Bitcoin hashing is one revenue line among several, and not necessarily the largest. For Bitcoin's network security, the implications of that transformation are still unfolding.