Publicly listed Bitcoin miners sold 32,000 BTC in Q1 2026 and signed more than $70 billion in cumulative AI and high-performance computing contracts. The sector's largest operators — Hut 8, TeraWulf, IREN, Core Scientific, and MARA Holdings — are converting mining facilities into GPU data centers...
"Bitcoin miners have energy available today. It's an easy pivot." — Fred Thiel, CEO, MARA Holdings
Publicly listed Bitcoin miners sold 32,000 BTC in Q1 2026 and signed more than $70 billion in cumulative AI and high-performance computing contracts. The sector's largest operators — Hut 8, TeraWulf, IREN, Core Scientific, and MARA Holdings — are converting mining facilities into GPU data centers for clients including Microsoft, CoreWeave, and undisclosed investment-grade tenants. Mining stocks have outperformed Bitcoin by roughly 70 percentage points year-to-date, reflecting a market that now prices these firms on contracted AI revenue, not hashrate.
The pivot is driven by a structural break in mining economics. Post-halving production costs for U.S.-listed miners exceed $80,000 per coin on a fully loaded basis, while Bitcoin trades near $67,000. At those margins, every megawatt redeployed from ASICs to NVIDIA GPUs represents immediate economic uplift. The result is the most consequential capital reallocation in crypto-mining history, with implications for Bitcoin's hashrate, network security, and the broader energy market.
The April 2024 halving cut the block subsidy from 6.25 BTC to 3.125 BTC, halving miners' guaranteed per-block revenue overnight. For publicly listed miners, the weighted average cash cost to produce one bitcoin rose to approximately $79,995 in Q4 2025, according to CoinDesk data. When hardware amortization, site maintenance, and corporate overhead are included, the fully loaded cost for many U.S.-based operations exceeds $100,000 per coin.
Bitcoin's spot price has hovered around $67,000 for much of 2026, placing the majority of publicly listed miners in negative margin territory on a per-coin basis. At current network difficulty, only operators with power costs below $0.05/kWh and hardware efficiency under 20 J/TH can produce bitcoin at a profit. According to Spark Research, a 1 EH/s operation at $0.03/kWh produces at roughly $20,000–$25,000 all-in cost; the same operation at $0.07/kWh approaches $45,000–$55,000.
The efficiency floor has risen sharply. Hardware running above approximately 25 J/TH now operates at a loss in most U.S. electricity markets. Modern ASICs have achieved sub-15 J/TH, representing a 7x improvement over eight-year-old equipment — but capital expenditure for fleet upgrades runs into the hundreds of millions, competing directly with AI infrastructure investment.
The scale of announced AI and HPC contracts across the public mining sector is unprecedented. Over $70 billion in cumulative contract value has been disclosed, transforming what were once pure-play mining companies into data center operators.
Hut 8 signed a 15-year, triple-net lease at its Beacon Point campus in Nueces County, Texas, covering 352 MW of IT capacity built to NVIDIA's DSX reference architecture. Base-term contract value: $9.8 billion, with options extending to $25.1 billion. The tenant is an undisclosed, investment-grade company. Hut 8 now holds 597 MW of contracted AI data center capacity worth approximately $16.8 billion in total. Shares rose 30% on announcement day in May 2026.
IREN secured a five-year, $9.7 billion GPU cloud services contract with Microsoft, providing access to NVIDIA GB300 GPUs at its Childress, Texas facility. IREN also signed a $5.8 billion equipment deal with Dell Technologies. The company reported Q1 FY26 revenue of $240.3 million, up 355% year-on-year, alongside $384.6 million in net income.
TeraWulf locked in $12.8 billion in contracted HPC revenue through deals with Google-backed Fluidstack and Core42 across sites totaling over 1 GW of available power. In Q1 2026, HPC leasing income reached $21 million versus $13 million from Bitcoin mining — the first quarter in which AI revenue exceeded mining revenue. Management has stated its intention to exit Bitcoin mining entirely by end of 2026.
Core Scientific holds approximately $10 billion in contracted revenue through CoreWeave partnerships spanning 590 MW under contract over twelve years. In Q1 2026, colocation revenue reached $77.5 million out of $115.2 million total — approximately two-thirds of group income. The company rejected a $9 billion CoreWeave acquisition bid, betting that its standalone value trajectory exceeds the offer.
MARA Holdings formed a joint venture with Starwood Capital Group to convert existing mining sites into AI-optimized data centers. CEO Fred Thiel described the pivot as risk-reduced infrastructure conversion rather than greenfield development.
To fund the capital-intensive transition, publicly listed miners collectively sold over 32,000 BTC in Q1 2026, a record quarterly sell-off. The proceeds are flowing directly into GPU procurement, facility retrofits, and power infrastructure upgrades.
Notable dispositions include:
The treasury liquidations represent a philosophical shift. For years, the "HODL" strategy — accumulating mined bitcoin on the balance sheet — was central to mining company investor narratives. That thesis has been abandoned in favor of contracted, dollar-denominated AI revenue. According to CoinDesk, miners are supplementing BTC sales with debt issuances to accelerate the transition timeline.
The market has endorsed the pivot decisively. While Bitcoin trades down approximately 20% year-to-date, mining equities have rallied sharply:
| Company | YTD Stock Return | Primary AI Partner | |---------|----------------:|-------------------| | TeraWulf | +73.6% | Fluidstack / Core42 | | Hut 8 | +67% | Undisclosed (inv. grade) | | Riot Platforms | +46% | Multiple | | Core Scientific | +40% | CoreWeave | | Applied Digital | +37% | Multiple | | IREN | +500% (trailing 12mo) | Microsoft | | Bitdeer | +5% | — |
The sector's 10 largest stocks all traded in positive territory YTD as of late May 2026. According to 24/7 Wall Street, Bitcoin mining ETFs were up over 50% while Bitcoin itself was down 7% — a divergence that "is the real crypto story of 2026."
The repricing reflects a fundamental shift in how investors value these companies: not on hashrate or BTC production, but on contracted kilowatt capacity, counterparty credit quality, and data center buildout timelines. A miner with 15-year, take-or-pay AI leases generates predictable cash flows that bear more resemblance to a REIT than a commodity producer.
Bitcoin's network hashrate briefly crossed 1 ZH/s (1,000 EH/s) in January 2026, setting multiple all-time highs with the 7-day moving average reaching 1.05–1.13 ZH/s. Since then, it has declined approximately 4% year-to-date as miners redirect capacity.
According to CoinDesk, Q1 2026 marked the first quarter-over-quarter hashrate decline in six years. Mining difficulty is projected to drop approximately 9% in the next adjustment around June 13, 2026, falling from roughly 139 trillion to 126 trillion, according to DEXTools data.
The difficulty reduction is the network's self-correcting mechanism: as miners leave, remaining participants face lower difficulty and improved economics. Private miners operating at lower cost structures — particularly those in Paraguay, Ethiopia, and other Global South locations with sub-$0.03/kWh hydropower — stand to benefit from the public miners' exit.
The core economic logic is straightforward: a megawatt of power allocated to NVIDIA GPUs running AI inference generates substantially higher and more predictable revenue than the same megawatt running SHA-256 ASICs.
According to S&P Global Market Intelligence, AI data center hosting contracts typically guarantee 5–15 year terms with investment-grade counterparties, at power densities of 40–80 kW per rack versus 5–10 kW for mining. The revenue per megawatt for AI hosting ranges from $1.5 million to $3.5 million annually, versus $0.3 million to $0.8 million for Bitcoin mining at current prices and difficulty — a 3x to 10x uplift depending on contract structure.
The IEA has warned that combined electricity demand from data centers, AI, and crypto could double by 2026. U.S. wholesale electricity prices are projected to rise 8.5% this year, further pressuring miners' already-thin margins and accelerating the reallocation to higher-value compute.
Over 56.7% of the Bitcoin network's energy mix now comes from sustainable sources, up from 34% in 2021. However, the miners exiting are disproportionately U.S.-based operators with access to grid power, potentially shifting the network's geographic distribution toward regions with cheaper, often renewable, stranded energy assets.
Publicly listed U.S. miners have accounted for over 40% of global hashrate. Their partial or full exit from mining raises questions about network security concentration. However, the analysis is not straightforward.
A declining hashrate does not automatically reduce security. Bitcoin's difficulty adjustment ensures block times remain near 10 minutes regardless of total hashpower. The relevant security metric is the cost to mount a 51% attack, which depends on both hashrate and the geographic/operational distribution of remaining miners.
Some analysts argue that reduced concentration among publicly listed U.S. companies could improve decentralization. If the hashrate migrates to a broader set of private operators across multiple jurisdictions, the network may become more resilient to regulatory or geopolitical disruption — even at a lower absolute hashrate.
The counterargument: a hashrate decline driven by economic capitulation, not technological improvement, signals weakening miner commitment. If Bitcoin's price does not recover to levels that make mining profitable at scale, the long-term security budget — funded by block rewards and transaction fees — may prove insufficient.
The public Bitcoin mining sector is undergoing a structural transformation, not a cyclical adjustment. The combination of halved block rewards, rising energy costs, and the availability of higher-margin AI workloads has created a one-way economic incentive to reallocate power capacity. Companies that entered 2024 as Bitcoin miners are exiting 2026 as data center operators.
For Bitcoin's network, the near-term effect is a modest hashrate decline and difficulty reduction that improves economics for remaining miners — a self-correcting mechanism working as designed. The longer-term question is whether transaction fees alone can sustain adequate security spending as block rewards continue to halve.
For investors, the mining sector's repricing demonstrates a market that has learned to value infrastructure optionality over single-asset exposure. The companies best positioned are those with grid-connected power capacity, regulatory relationships, and the engineering talent to serve investment-grade AI clients. Bitcoin mining, for many of these firms, has become a transitional activity rather than a core business.