Publicly listed Bitcoin miners sold more than 32,000 Bitcoin in the first quarter of 2026, a record liquidation that underscores how the sector’s largest operators are redirecting billions of dollars from mining into artificial intelligence and high‑performance computing. The accelerated shift is arriving as mining profitability sits near cyclical lows and the economic model for validating Bitcoin faces renewed pressure after the April 2024 halving.

Market Impact

The scale of first‑quarter selling highlights how balance sheets are being repurposed to fund new lines of business. Public mining firms offloaded more Bitcoin in the first three months of 2026 than they did across all of 2025, and the tally dwarfed the roughly 20,000 Bitcoin sold during the market turmoil surrounding the Terra‑Luna collapse in the second quarter of 2022. On‑chain data from CryptoQuant show miner reserves trending lower throughout the cycle, with leading operators increasingly relying on their Bitcoin treasuries for liquidity rather than treating them as strategic, long‑term holdings.

Since the start of the current cycle, miners have recorded a net sell of 61,000 BTC. Marathon Digital led the disposals, unloading more than 13,000 BTC and dropping out of the top three Bitcoin holders. Other sellers included Cango, which sold 2,000 Bitcoin for roughly $143 million to eliminate Bitcoin‑backed debt, Core Scientific, which sold around 1,900 Bitcoin in January to raise $175 million, and Riot Platforms, which sold 4,026 BTC. Collectively, these moves reflect a sector using liquid reserves to navigate rising costs and shifting strategic priorities.

Post‑Halving Economics

The April 2024 halving, which reduced block rewards from 6.25 BTC to 3.125 BTC, reset the sector’s revenue baseline and amplified vulnerability to spot‑price swings. Since then, mining economics have been defined by persistent pressure. CoinShares’ head of research, James Butterfill, noted that the weighted average cash cost for public operators to produce a single Bitcoin rose to nearly $80,000 in the final quarter of 2025.

On the revenue side, hashprice — the expected revenue per unit of computing power — fell to between $28 and $30 per petahash per second per day in the first quarter of 2026, a level near historical lows. With transaction fees contributing less than 1% of total block rewards, miners remain heavily dependent on Bitcoin’s spot price. At around $77,000 today, Bitcoin trades well below its cycle peak of approximately $126,000 reached in October 2025. Combined with elevated electricity costs and heavier debt service, cash flows are under strain, pushing executives to look beyond pure mining revenues.

AI Integration

That search is increasingly landing on AI and high‑performance computing. Unlike the volatile, short‑cycle dynamics of Bitcoin mining, AI data center services are typically sold under stable, multi‑year contracts with counterparties such as Google, Microsoft, and Anthropic. Equity markets have rewarded this repositioning. Mining companies that set AI revenue targets of 80% or more have seen their share prices rise sharply over the past two years, securing higher market multiples than peers that remain predominantly focused on mining.

Butterfill estimates that public miners could derive up to 70% of their revenues from AI by the end of this year, up from roughly 30% today. Across the public mining cohort, more than $70 billion in cumulative AI and high‑performance computing contracts have been announced, signaling a decisive capital reallocation away from next‑generation ASIC upgrades and toward data‑center‑style infrastructure. Operators including TeraWulf, IREN, and Cipher have taken on billions in collective debt to finance these buildouts, reflecting the appeal of the underlying unit economics: while electricity accounts for roughly 40% of Bitcoin mining revenue, energy costs for AI cloud operators leasing high‑powered chips sit in the low single digits.

This divergence in cost structures helps explain why treasuries are being monetized and why new investment is flowing into facilities designed for GPUs and other accelerators rather than for specialized mining hardware. Related analysis also suggests that, for the leading public miners, potential Bitcoin earnings in the current environment could still exceed long‑term AI contract revenue on an absolute basis — a reminder that the balance between these businesses is dynamic and will continue to shape strategy.

Technology Use Case

The physical footprint that miners control — power access, cooling capacity, and industrial‑scale campuses — maps closely onto the requirements of AI compute. As a result, repurposing or expanding existing infrastructure for training and inference workloads can be faster than building greenfield sites. This flexibility allows firms to allocate capacity between AI and mining as economics shift, using AI as a steady baseload business while keeping optionality to mine when margins improve.

Industry Response

The reallocation of capital has sparked debate about the long‑term security of the Bitcoin network. Charles Edwards of Capriole Investments warns that the average Bitcoin revenue share among the top public miners could fall to about 30% within three years, interpreting the trend as a potential reduction in the sector’s energy commitment to the network. Researcher Paul Sztorc sees cultural signs of the transition as well, pointing to mining publications that now emphasize broader energy themes and conferences that have replaced dedicated mining stages with energy‑focused programming.

Others argue the protocol is designed to adapt. Blockstream CEO Adam Back highlights Bitcoin’s self‑adjusting difficulty. When computing power exits, difficulty declines, improving margins for remaining miners until economic equilibrium reemerges. He characterizes the migration as an arbitrage that naturally balances AI and mining workloads and notes a “positive reflexivity,” where higher margins reduce the need for miners to sell Bitcoin to cover operating costs.

On‑chain analyst James Check offers a similar perspective, describing high turnover as an intended feature of difficulty adjustment. In his framing, diversified infrastructure firms “buy power and compute” and allocate it to the best‑paying use case at any given moment. In that model, AI provides a stable baseline, while Bitcoin mining functions as a flexible, intermittently deployed tool that can complement grid dynamics.

Looking Across the Halving Cycle

The network recently crossed block 945,000 in April 2026, placing it in the second half of the current halving epoch. Hashrate Index argues that the next two years, leading into the 2028 halving, will test how effectively Bitcoin’s self‑correcting incentives can compete with the pull of Wall Street capital into AI. The questions ahead are structural: whether Bitcoin’s spot price can rise enough to clear near‑record cash production costs, whether transaction fees will remain a negligible share of total revenue, and whether the current pace of treasury liquidations can continue without damping asset prices.

Another unknown is where network computing power ultimately stabilizes once higher‑cost operators exit. However these variables resolve, the identity of the public mining industry is already shifting. By 2027, many of the companies that helped industrialize Bitcoin validation may operate primarily as diversified energy and high‑performance computing platforms, with residual exposure to the asset that launched them — and with capital allocation increasingly governed by the economics of AI.