Public Bitcoin miners sold a record 32,000 BTC in Q1 2026—more than the combined total for all four quarters of 2025. The liquidation funds a structural identity shift: CoinShares projects up to 70% of listed miner revenue will derive from AI and high-performance computing infrastructure by year-...
"Bitcoin mining investment doesn't make a lot of sense at current hashprices compared to the returns available in AI infrastructure." — CleanSpark executive, Q1 2026 earnings call
Public Bitcoin miners sold a record 32,000 BTC in Q1 2026—more than the combined total for all four quarters of 2025. The liquidation funds a structural identity shift: CoinShares projects up to 70% of listed miner revenue will derive from AI and high-performance computing infrastructure by year-end, up from approximately 30% in Q4 2025. The Bitcoin mining industry, two years past the April 2024 halving, is no longer primarily a Bitcoin mining industry.
The numbers frame the crisis. Hashprice—daily revenue per petahash of deployed compute—collapsed to $27.89/PH/s/day in early March, its lowest reading since 2018. Weighted average cash cost per BTC among public miners rose to $79,995 in Q4 2025, per CoinShares. With Bitcoin trading at $78,281 on May 3, 2026, the margin between production cost and market price has effectively vanished for all but the most efficient operators. Network difficulty has declined in six of its nine adjustments in 2026, with the most recent 2.3% drop on May 1 pushing difficulty to 132.47 trillion as hashrate slipped below 1 ZH/s.
Bitcoin's network hashrate peaked above 1 ZH/s (1,000 EH/s) in late 2025, briefly touching that milestone before retreating. As of May 3, 2026, CoinWarz reports the network at approximately 997 EH/s, though 30-day averages have stabilized lower at roughly 920-1,004 EH/s depending on the measurement window.
Difficulty adjustments in 2026:
Hashprice trajectory:
The hashprice recovery in April owes more to difficulty relief than price appreciation. Bitcoin moved from ~$69,200 in mid-March to $78,281 by May 3, while difficulty fell, widening the spread for surviving operators. CoinShares estimates 15-20% of the global mining fleet is unprofitable at hashprices below $30/PH/s/day, predominantly operators running pre-S19 XP hardware at power costs above $0.06/kWh.
According to CoinShares' Q1 2026 Bitcoin Mining Report, the weighted average cash cost to produce one BTC among listed miners reached $79,995 in Q4 2025. This figure includes direct electricity costs, hosting fees, and site-level operating expenses but excludes depreciation, SG&A, and interest.
Cost breakdown by efficiency tier:
| Tier | Hardware | Power Cost | Est. Cost/BTC | Status | |------|----------|------------|---------------|--------| | Tier 1 | S21/S21+ (sub-15 J/TH) | <$0.05/kWh | $34,000-$43,000 | Profitable | | Tier 2 | S19 XP (21.5 J/TH) | $0.05-0.06/kWh | $55,000-$68,000 | Marginal | | Tier 3 | S19j Pro (29.5 J/TH) | $0.06-0.08/kWh | $74,000-$88,000 | Underwater | | Tier 4 | Legacy (<S19) | >$0.08/kWh | >$100,000 | Shutdown |
In March 2026, CoinDesk reported miners were "losing $19,000 on every BTC produced" when Bitcoin traded near $69,200 and average all-in production costs sat at approximately $88,000. The situation has improved marginally with BTC's recovery to $78,000, but Tier 3 operators remain cash-negative.
The electricity-only breakeven—the point where power costs alone equal the value of mined BTC—sits near $74,000 for mid-tier operators. With U.S. industrial electricity averaging $0.07-0.08/kWh in Q1 2026, only miners with access to sub-$0.05/kWh power maintain meaningful operating margin.
Public miners collectively sold 32,000 BTC in Q1 2026, per TheMinerMag data cited by CoinDesk. This exceeds combined sales for all of 2025 and represents the most aggressive treasury drawdown since the industry's public market emergence in 2020-2021.
Notable sales:
The selling serves a dual purpose: covering operating losses at compressed margins and funding capital expenditure for AI/HPC infrastructure buildout. This is not panic selling—it is strategic asset reallocation from a depreciating revenue stream (BTC mining at negative margins) toward higher-return infrastructure contracts.
MARA Holdings' position illustrates the tension. The company held 44,893 BTC as of December 31, 2025, and operated 66.4 EH/s of hashrate. Despite this scale, MARA stock is down 28% year-to-date—the only major miner in negative territory. Its March sale of 15,133 BTC suggests a pivot toward deploying capital elsewhere, including its acquisition of a 64% stake in Exaion, a French AI data center company.
The aggregate value of AI and HPC contracts signed by former Bitcoin miners now exceeds $70 billion. CoinShares projects these contracts could deliver up to 70% of listed miner revenue by year-end 2026, fundamentally reshaping the sector's identity.
Key contracts and deals:
| Company | Deal | Value | Duration | Partner | |---------|------|-------|----------|---------| | Core Scientific | AI data center hosting | $10.2B | 12 years | CoreWeave | | Hut 8 | AI data center | $7.0B | 15 years | Fluidstack (Google-backed) | | Iris Energy | GPU cloud expansion | $3.4B annualized target | Ongoing | Multiple | | Core Scientific | Bond sale for AI buildout | $3.3B | — | Public markets |
Core Scientific represents the most complete transformation. The company is converting a 300 MW Bitcoin mining operation in Pecos, Texas, into a 1.5 GW AI data center campus. AI colocation already contributed 39% of quarterly revenue as of late 2025. The proposed $9 billion all-stock acquisition by CoreWeave (announced July 2025, offering 66% premium) would complete the metamorphosis from miner to pure-play AI infrastructure.
Hut 8 has secured what analysts call the most credible "real revenue" deal in the sector—a $7 billion contract with Google-backed Fluidstack for a 245 MW AI facility over 15 years. The company partnered with Anthropic for AI workloads.
Iris Energy (IREN) leads on scale, with a $14 billion market cap and a target of 140,000 NVIDIA GPUs across a 4.5 GW development pipeline. The company targets $3.4 billion in annualized AI cloud revenue by end-2026.
CleanSpark remains a notable holdout, prioritizing mining efficiency over AI—though it recently unveiled plans for AI data center deployment using existing land and compute infrastructure. CleanSpark stock is up 53% year-to-date, suggesting the market rewards both strategies when executed well.
Year-to-date stock performance (as of May 2, 2026):
| Company | YTD Return | Primary Strategy | |---------|-----------|-----------------| | CleanSpark (CLSK) | +53% | Mining-focused, nuclear-powered | | Iris Energy (IREN) | +42%* | AI-first, massive GPU pipeline | | Core Scientific (CORZ) | +31%* | Full AI pivot, CoreWeave deal | | Hut 8 (HUT) | +28%* | AI contracts, hybrid model | | MARA Holdings (MARA) | -28% | Large BTC treasury, slow pivot |
*Approximate figures based on available reporting.
The market is pricing AI optionality at a premium. Mining stocks with credible AI contracts have outperformed Bitcoin itself in 2026, while companies perceived as "just miners" trade at discounts to their BTC holdings. This creates an unusual dynamic where the mining sector's equity performance has decoupled from Bitcoin's price action.
For Bitcoin's network security: The hashrate decline from >1 ZH/s to sub-1 ZH/s represents a 5-10% contraction in network security. However, difficulty auto-adjustment ensures block production remains near 10-minute targets. The risk is concentration: as marginal miners exit, the remaining hashrate concentrates among fewer operators, potentially increasing 51% attack vulnerability at the margin.
For BTC supply dynamics: The 32,000 BTC quarterly liquidation from miners adds persistent sell pressure. At current production rates (~450 BTC/day from block rewards), miners are selling approximately 2.4x their production—drawing down accumulated treasuries. This selling pressure dissipates as treasuries deplete, but in the near term represents a headwind equivalent to ~$830M in quarterly sell flow at current prices.
For AI infrastructure: Former miners bring purpose-built power infrastructure, cooling systems, and land to the AI buildout. Their existing 5-10 GW of aggregate power capacity addresses one of AI's binding constraints. The conversion economics are favorable: AI hosting generates $300-500/kWh annually versus $50-100/kWh for Bitcoin mining at current difficulty.
For the mining identity: The sector's taxonomy is shifting. "Bitcoin miner" as a category may not exist in its current form by 2027. What replaces it is a hybrid infrastructure operator that mines BTC as a byproduct of excess capacity rather than a primary business line.
The Bitcoin mining industry is undergoing a structural metamorphosis rather than a cyclical downturn. Previous halving cycles produced temporary margin compression followed by recovery as BTC price appreciated. This cycle differs because an alternative revenue stream—AI infrastructure—offers superior risk-adjusted returns on the same underlying assets (power, land, cooling). When AI hosting generates 3-5x the revenue per kilowatt-hour of Bitcoin mining at current difficulty, the rational economic actor diverts capacity.
The 32,000 BTC quarterly sell-off is the financial expression of this calculation. Miners are not capitulating—they are reallocating. The Bitcoin network will maintain security through difficulty adjustments and the remaining operators who retain structural cost advantages (stranded energy, state-backed mining, latest-generation hardware at sub-$0.05/kWh). But the publicly listed "mining sector" as investors have known it since 2020 is becoming an AI infrastructure sector that happens to mine Bitcoin on the side.
CoinShares' projection—70% AI revenue by end-2026—implies the transition's midpoint has already passed. The endpoint is a sector where "Bitcoin mining" is a line item, not a business description.