What the AI Pivot Means for Bitcoin Miners — and Bitcoin

North American listed miners are responding to challenging post-halving economics by pivoting toward AI infrastructure — but not uniformly, and not without risk. The shift is splitting the sector into different kinds of companies. Bitcoin, meanwhile, looks likely to absorb the change largely as designed.

One of the biggest stories of 2026 so far has been the shift by North American listed bitcoin miners into Artificial Intelligence (AI) infrastructure. MARA and CleanSpark became the latest examples on May 11, 2026, both posting heavy losses for the quarter ended March 31, while elaborating on plans to expand their focus on AI and high-performance computing (HPC).

Often framed as a story about companies abandoning Bitcoin en masse, the reality is more nuanced. It does, however, raise a key question: is scarce power better used mining bitcoin, or leased to an AI industry willing to pay for long-term capacity?

The answer is different for every company. Some miners are moving decisively away from Bitcoin. Others remain mining-heavy, but are building optionality around power, land and data-centre infrastructure. One thing the Q1 reporting cycle has made clear, however, is that the “bitcoin miner” label now covers companies with very different underlying businesses.

The Economics Behind the Shift

The most obvious reason for the pivot among many miners is simple economics. After the April 2024 halving, miners continued to compete for 3.125 BTC per block as part of a network that grew throughout much of 2025. According to CoinShares’ Q1 2026 Mining Report, by October 2025, global hashrate had reached an all-time high of roughly 1,160 EH/s. Despite pulling back into late 2025 and early 2026, competition remained fierce enough to push hashprice, the daily revenue generated per petahash, to roughly $29/PH/s in Q1 2026. 

The weighted average cost to produce one BTC among listed miners meanwhile sat at roughly $80,000 in Q4 2025, with 15 to 20 percent of the global fleet estimated to be operating at a loss. All of this against the backdrop of rapid growth in the AI sector.

Revenue from mining is notoriously volatile, tied directly to BTC price, network difficulty and energy costs. The AI and HPC industries, in contrast, offer what — at least for now — appears to be higher and more predictable long-term revenue that lenders are more inclined to finance.

A significant portion of the mining sector is built around assets that AI infrastructure providers need and cannot easily replicate themselves.

The overlap among the two industries includes large power purchase agreements, grid-connected land, and facilities that can both run energy-hungry hardware and, in some cases, generate their own power.

For miners caught between rising production costs and falling hashprice, the temptation to shift operations away from thinner, more uncertain mining returns is easy to understand.

Not All Miners Are Making the Same Bet

For some miners, the move to make bitcoin mining secondary — and potentially exit it entirely — is already underway.

Core Scientific, for example, reported that its bitcoin mining segment ran at a negative gross margin in Q1 2026. Its colocation business, by contrast, remained highly profitable. While the company has not exited mining entirely, it has signalled it is no longer a priority, describing self-mining as a way to help offset power costs while it scales towards almost 600 MW of AI capacity.

Hut 8, meanwhile, has moved to separate its bitcoin mining operations through American Bitcoin Corp, repositioning Hut 8 itself as an energy infrastructure platform. At the more decisive end of the spectrum, Keel Infrastructure, formerly Bitfarms, said in February 2026 that it is no longer a Bitcoin company, while Cipher’s CEO told investors on an earnings call in May that bitcoin mining will cease to be part of the company’s story by 2030.

MARA Holdings sits in a more ambiguous middle ground. Its planned acquisition of Long Ridge Energy & Power gives it control of a 505 MW gas-fired power plant and more than 1,600 acres of industrial land in Ohio, creating a clearer path into AI, HPC and broader digital infrastructure. Operationally, however, it remains one of the largest bitcoin miners in the market, with 72.2 EH/s of energised hash rate in Q1, up 33 percent year-on-year. Riot is somewhat harder to read. It generated $111.9 million of bitcoin mining revenue in Q1, far above its reported $33.2 million in data-centre revenue. Most of that figure, however, was reimbursement for construction work rather than recurring lease income. Its core business remains bitcoin mining, even if its infrastructure strategy is clearly changing.

Widely seen until recently as one of the few remaining pure-play listed bitcoin miners in North America, CleanSpark reported fiscal Q2 results in May 2026, showing it had increased its average monthly hashrate by 18 percent year-on-year, despite an almost 25 percent fall in revenue. At the same time, it has doubled its megawatts under contract over the past year, with much of that capacity now earmarked for AI infrastructure.

Trading One Risk for Another

The AI contract backlogs being announced across the sector are large and long-dated. Earning them requires companies to spend heavily upfront, complete construction on schedule, secure power delivery and keep customers committed to leases that can run for well over a decade. A contract backlog is a claim on future execution, not a guarantee.

These commitments also reduce future flexibility. A miner that commits scarce power capacity to long-term AI leases cannot easily switch it back to Bitcoin if hashprice recovers or bitcoin rallies. 

For companies moving hardest into AI, the pivot may solve today’s margin problem while giving up tomorrow’s mining optionality, just as weaker competitors exit and the economics for remaining miners improve.

The financing required to fund that buildout is also substantial. Several miners have raised billions in project debt against future lease revenue, helped by credit support from hyperscalers including Google and Microsoft. That makes institutional financing easier, but it does not absorb the operational challenge of building hyperscale data-centre infrastructure, where specialised equipment, power delivery, construction timelines and customer performance all become sources of risk.

The trade comes down to hashprice volatility versus infrastructure execution risk. For an operator currently mining at a cash loss, that may be rational. But the shift exchanges one set of risks for another, with no guarantee the new set is smaller.

Bitcoin Is Working as Designed

The obvious question is whether public miners shifting capacity away from Bitcoin weakens the network in any meaningful way. A sharp, sustained hashrate decline would reduce Bitcoin’s security margin by making the network cheaper to attack.

The go-to comparison is China’s 2021 mining ban, when a much larger share of global hashrate came offline in a short period. In the end, blocks slowed, Bitcoin’s difficulty adjusted and mining activity migrated elsewhere. The episode was disruptive for miners, but not existential for the network.

The more important question is who replaces uneconomic hashpower. CoinShares argues resilience has been supported by state-backed miners, private operators with cheap or stranded power and ASIC manufacturers running unsold inventory through their own facilities. 

Hashrate may also be becoming more geographically dispersed, with Paraguay, Ethiopia and Oman having recently entered the global top 10. That could reduce one form of concentration, though opaque or state-linked replacement hashpower brings its own risks.

Listed miners are businesses like any other, and businesses reallocate capital when the opportunity cost changes. Throughout Bitcoin’s history, miners have entered, exited and relocated. Difficulty has adjusted, and hashpower has followed the cheapest and most durable sources of energy.

If some North American public miners decide their power is worth more serving AI, that will clearly change who earns future block rewards. It may also change how public miner equities trade.

What the shift shows is not the failure of bitcoin mining. It is the network is responding exactly as designed.