The Mining Industry Just Stopped Looking Like a Mining Industry

The latest difficulty adjustment was a marker. On May 15, 2026, Bitcoin mining difficulty dropped by 7.76%, the second-largest decline of the year. The fall came as several of the largest public miners curtailed operations, sold inventory, and accelerated a transition that has reshaped the entire industry in less than 18 months. The publicly listed mining sector is no longer primarily a Bitcoin business. It is becoming an AI infrastructure business with a Bitcoin treasury attached.

This piece is about why that happened, what the numbers look like, and what it means for the security and economics of the Bitcoin network itself.

The Hashprice Collapse

Hashprice — the daily revenue, in dollars, generated by one petahash per second of mining capacity — is the single most important variable in mining economics. In May 2026 hashprice has been trading in the $34 to $35 per PH/s/day range. CoinShares reports the bottom touched roughly $29 in February. By any historical comparison, this is a five-year low, and it sits well below the breakeven for many fleets that were profitable before the April 2024 halving.

Two forces compressed the metric. The halving cut the block subsidy in half, dropping daily issuance from approximately 900 BTC to 450 BTC. Then the network's hashrate kept climbing — pushing close to 1 ZH/s through 2025 and 2026 — as miners deployed new-generation ASICs ordered before the halving. Higher network hashrate divided over the same number of bitcoin per day equals less revenue per unit of compute. The math is unforgiving.

The result is that the average cost to produce one bitcoin reached about $49,645 in fiscal year 2025, up 54% from $32,216 in fiscal year 2024. CoinShares estimates that roughly 20% of the world's legacy mining fleet is currently operating at a loss at current prices.

What AI Hosting Pays Compared to Mining

Where mining squeezes margins, AI hosting expands them. The contracts public miners have signed over the past 12 months pay multiples of what their mining revenue produced on the same megawatt of capacity.

The headline example is Riot Platforms. Riot reported $167.2 million in Q1 2026 revenue. Inside that number, $33.2 million came from data center operations, a line that did not meaningfully exist a year earlier. The driver is a contract with AMD for high-performance computing capacity. AMD recently exercised an option for an additional 25 MW, taking the total contracted capacity to 50 MW of critical IT load. Riot sold 3,778 BTC during the same window for roughly $289.5 million, monetizing the treasury to fund the AI build-out.

Marathon Digital has moved faster on the operational side. The company offloaded 15,133 BTC in March, ending Q1 2026 with 52,850 BTC in treasury, worth roughly $3.4 billion. Rather than concentrating all capacity in a few flagship sites, Marathon is rolling out smaller containerised installations of about 10 MW each, deployed at the edge of energy networks where power is cheap and grid interconnects are faster to obtain. The structure is identical to the modular AI build-outs that hyperscalers have used to expand inference capacity in 2025 and 2026.

Core Scientific has been the most aggressive in repositioning balance-sheet capacity into AI hosting, which is part of the reason this month's difficulty adjustment came in negative. When a company that runs hundreds of megawatts of capacity reroutes contracted power from SHA-256 to GPUs, the global hash rate visibly responds.

CoinShares' research team projects that some publicly listed miners could derive up to 70% of total revenue from AI hosting by the end of 2026. That projection would have looked far-fetched 18 months ago. Today it is the consensus.

Why the Switch Works Economically

Bitcoin mining and AI hosting share three primary inputs: electrical power, real estate, and cooling. They differ on the revenue side.

A Bitcoin miner sells a commodity (BTC) into a volatile spot market, with revenue tied to a price the operator does not control and a subsidy that halves every four years. An AI host sells contracted megawatts to a single enterprise customer on multi-year terms, often with take-or-pay clauses. The revenue per MW is two to three times higher than mining at current hashprice, and the cash-flow profile is contractual rather than market-driven.

Conversion is not free. Mining sites need GPU-grade cooling, network connectivity, redundant power, and physical-security upgrades that mining did not require. Capex per MW for AI conversion has been running between $5 million and $9 million depending on tier and customer mix. But the payback math is favorable when the alternative is sub-economic SHA-256 hosting on aging ASIC fleets.

The strategic logic is also defensive. Public miners that do not capture AI revenue this cycle risk losing access to the cheap power contracts that defined their economic edge in the first place. AI customers are willing to pay above-market rates for energy, which makes mining-only operators a less attractive counterparty to power utilities.

What This Does to Bitcoin Network Security

The honest answer is: not much, yet.

Mining difficulty adjusts every 2,016 blocks to keep block times near the 10-minute target. When operators shut off rigs, difficulty falls and the remaining miners earn more per unit of hash. That self-regulating loop is why even a 7.76% drop in difficulty does not damage Bitcoin's settlement assurance. The network simply finds a new equilibrium.

The longer-term question is whether the marginal miner — the small or mid-sized operator that does not have an AI hosting option — can survive. If the AI pivot continues to draw hashrate away from listed miners, the share of network hashrate held by private operators, sovereign-aligned facilities (especially in the Middle East), and stranded-energy projects will grow. None of those operators face the same quarterly-earnings pressure that has forced public miners to switch.

Geographic concentration is a related concern. CoinShares' Q1 2026 report flagged that the share of hashrate produced in North America has begun to flatten as new capacity gets prioritized for AI rather than SHA-256. Whether the next wave of growth comes from Latin America, the UAE, or older Asian markets is the more interesting structural question for the second half of 2026.

The Investor View

For shareholders of public miners, the AI pivot has been a clear positive. Riot's data center revenue has stabilized quarterly earnings. Marathon's BTC treasury, currently worth around $4 billion if marked at $79,000 per coin, gives the company strategic optionality that a pure mining shop would not have. The operating risk is execution: building reliable AI hosting infrastructure is a different muscle from running a fleet of ASICs.

For Bitcoin holders, the question is more nuanced. A miner base that is less dependent on BTC price is a more stable miner base, which is good for network security through cycles. But it also means that miner selling — historically a meaningful source of overhead supply — is becoming less price-sensitive. Miners no longer need to sell BTC at any specific price to fund operations because their cash flow is increasingly contractual.

Where the Story Goes From Here

Three milestones will define the next 12 months.

The first is the AMD and similar hyperscaler contracts that miners have already signed but not yet built out. Capacity that is contracted but not energized accounts for several hundred megawatts across the public-miner cohort. Each of those builds completing on schedule converts a press release into recurring revenue.

The second is the difficulty trajectory. A second consecutive negative adjustment would suggest the network is still working through the AI-driven curtailment. A return to positive adjustments would suggest the equilibrium has been reached and the remaining miners are once again competing for share.

The third is the BTC price itself. If Bitcoin re-rates higher in the second half of 2026, hashprice climbs mechanically and the AI pivot pace could slow. If BTC remains range-bound at $75,000-$85,000, the pivot accelerates and the publicly listed mining sector continues to look more like an AI hosting industry that happens to mine Bitcoin on the side.

Bottom Line

The AI pivot is not a passing rotation. It is a structural reallocation of capital and capacity that began the day the April 2024 halving cut block rewards in half and gained momentum as enterprise GPU demand ran ahead of hyperscaler build-out timelines. For Bitcoin, the change is neutral to mildly positive in the medium term. For the publicly listed miners, it is the difference between a structurally unprofitable business and a real one.

Frequently Asked Questions

What is hashprice and why does it matter?

Hashprice is the daily revenue, expressed in US dollars, that one petahash per second of mining capacity earns at current Bitcoin price and difficulty. It is the single best measure of mining profitability. As of May 2026, hashprice sits at roughly $34 to $35 per PH/s, a five-year low.

How does AI hosting compare to Bitcoin mining for a public miner?

AI hosting pays two to three times more per megawatt of power than mining at current hashprice, and the revenue is contractual rather than spot-market-dependent. The trade-off is higher capital expenditure per megawatt for cooling, networking, and security upgrades.

Are companies like Marathon and Riot exiting Bitcoin mining entirely?

Not yet. Both companies continue to mine and both hold large BTC treasuries (Marathon roughly 52,850 BTC; Riot has been selling more aggressively to fund its AI build-out). CoinShares projects that some public miners could earn up to 70% of total revenue from AI hosting by the end of 2026, but mining still contributes a meaningful share.

Does the miner AI pivot threaten Bitcoin's network security?

Not directly. Bitcoin's difficulty adjustment mechanism rebalances the network every 2,016 blocks, so when miners switch off rigs, remaining miners earn more per unit of hashrate and the system finds a new equilibrium. The more interesting structural question is whether private operators, sovereign-aligned facilities, and stranded-energy projects take a larger share of hashrate going forward.

Could miners come back to Bitcoin mining if prices rise?

Yes. Hashprice is a function of both Bitcoin price and network difficulty. If BTC re-rates significantly higher in the second half of 2026, the economics flip in favor of mining and some capacity could rotate back. But once a site is converted to AI hosting with multi-year customer contracts, the switching cost back to mining is substantial.

External References

  • [CoinShares Bitcoin Mining Report Q1 2026](https://coinshares.com/insights/research-data/bitcoin-mining-report-q1-2026/)
  • [Bitcoin Mining Difficulty Drop May 2026: AI Pivot & Operator Guide](https://millionminer.com/news/bitcoin-mining-difficulty-drop-ai-pivot)
  • [Riot Platforms Stock Forecast 2026 — MEXC](https://www.mexc.com/news/1049506)
  • [Bitcoin Security Risk: Miners Pivot to AI as Mining Difficulty Drops 7.76%](https://www.techi.com/bitcoin-miners-ai-pivot-security-risk/)

*Disclaimer: This article is for informational purposes only and does not constitute investment advice. Cryptocurrency markets are highly volatile and investors can lose part or all of their capital. Always conduct your own research and consult a qualified financial advisor before making investment decisions.*