For two centuries, factories chased cheap hands and dense ports. Today, miners roll into windy plateaus and hydro spillways, asking a simpler question: where are the cheapest wasted watts?
When computing can move to energy rather than energy to people, the map tilts.
Heavy industry has always chased cheap energy, but it still needed bodies and ships. The novelty with Bitcoin (BTC) is how completely labor, logistics, and physical product have dropped out of the siting equation.
A mining plant can be one warehouse, a dozen staff, a stack of ASICs, and a fiber line. Its output is pure block rewards, not a bulky commodity that must be shipped. That lets miners plug into genuinely stranded or curtailed energy that no conventional factory would bother to reach, and to rush when policy or prices change.
Bitcoin isn’t the first energy-seeking industry, but it is the first large industry whose primary location bid is “give me your cheapest wasted megawatt, and I’ll show up,” with labor nearly irrelevant.
Curtailment creates a new subsidy
CAISO curtailed about 3.4 TWh of utility-scale solar and wind in 2023, up roughly 30% from 2022, and saw more than 2.4 TWh curtailed in just the first half of 2024 as mid-day generation routinely overshot demand and transmission limits.
Nodal prices often go negative: generators pay the grid to take their electricity because shutting down is costly, and they still want renewable tax credits.
Miners show up as a strange new bidder. Soluna builds modular data centers at wind and solar projects that soak up power the grid cannot absorb. In Texas, Riot earned about $71 million in power credits in 2023 by curtailing during peak demand, often more than offsetting the BTC they would have mined.
In 2024, the Bitcoin mining firm turned curtailment into tens of millions of dollars of credits, and in 2025, it is on track to beat that, with more than $46 million of credits booked in the first three quarters alone.
A 2023 paper in Resource and Energy Economics models Bitcoin demand in ERCOT and finds that miners can increase renewable capacity but also emissions, with much of the downside mitigated if miners operate as demand-response resources.
Curtailment and negative pricing are a de facto subsidy for anyone who can show up exactly where and when power is cheapest, and mining is architected to do that.
Hash rate moves faster than factories
Miners used to seasonally migrate within China seasonally, chasing cheap wet-season hydropower in Sichuan and then shifting to coal regions like Xinjiang when the rains ended.
When Beijing cracked down in 2021, that mobility went global: US hash-rate share jumped from single digits to roughly 38% by early 2022, while Kazakhstan’s share spiked to around 18% as miners lifted whole farms and re-planted them in coal-heavy grids.
For the past year, US-based mining pools have mined over 41% of Bitcoin blocks.
Reuters recently reported that China’s share has quietly rebounded to around 14%, concentrated in provinces with surplus power.
ASICs are container-sized, depreciate in two to three years, and produce the same virtual asset regardless of where they sit. That lets hashrate slosh across borders in a way steel mills or AI campuses can’t.
When Kentucky exempts mining electricity from sales tax, or Bhutan offers long-term hydropower contracts, miners can pivot in months.
A programmable knob and wasted-watts frontier
ERCOT treats specific large loads as “controllable load resources” that can be curtailed within seconds to stabilize frequency.
Lancium and other mining facilities brand themselves as CLRs, promising to ramp down almost instantly when prices spike or reserves thin. Riot’s July and August 2023 reports read like grid-services earnings releases, with millions in power and demand-response credits booked alongside far fewer self-mined coins during heat waves.
The OECD and national regulators now discuss Bitcoin as a flexible load that can either deepen renewable penetration or crowd out other uses.
Miners bid on interruptible power at rock-bottom rates, grid operators gain a buffer they can call on during tight supply, and the grid absorbs more renewable capacity without overbuilding transmission.
Bhutan’s sovereign wealth fund and Bitdeer are building at least 100 MW of mining powered by hydropower as part of a $500 million green-crypto initiative, monetizing surplus hydro and exporting “clean” coins. Officials reportedly used crypto profits to pay government salaries.
In West Texas, wind and solar fleets run into transmission bottlenecks, producing curtailment and negative prices.
That is where many US miners have situated, signing PPAs with renewable plants to take capacity that the grid cannot always absorb. Crusoe Energy brings modular generators and ASICs to remote oil wells, using associated gas that would otherwise be flared.
Miners cluster where three conditions overlap: energy is cheap or stranded, transmission is constrained, and local policy welcomes or ignores them. Bitcoin mines can reach sites that a workforce-intensive industry never could.
AI adopts the playbook, with limits
The US Department of Energy’s Secretary’s Energy Advisory Board warned in 2024 that AI-driven data center demand could add tens of gigawatts of new load. It stressed the need for flexible demand and new siting models.
Companies like Soluna now pitch themselves as “modular green compute,” toggling between digital assets and other cloud workloads to monetize curtailed wind and solar.
China’s new underwater data center off Shanghai runs roughly 24 MW, almost entirely on offshore wind, with seawater cooling.
The friction comes from latency and uptime SLAs. A Bitcoin miner can tolerate hours of downtime and seconds of network lag.
An AI inference endpoint serving real-time queries cannot. That will keep tier-one AI workloads near fiber hubs and major metros, but training runs and batch inference are already candidates for remote, energy-rich sites.
El Salvador’s proposed Bitcoin City would be a tax-haven city at the base of a volcano, where geothermal power would feed Bitcoin mining, with Bitcoin-backed bonds funding both the town and miners.
Whether or not it gets built, it shows a government pitching “energy plus machines” rather than labor as the anchor. Data-center booms in the Upper Midwest and Great Lakes draw hyperscalers with cheap power and water despite limited local labor.
Bhutan’s hydropower-backed mining campuses sit far from major cities.
The civic fabric is thin. A few hundred high-skill workers service racks and substations. Tax revenue flows, but job creation per megawatt is minimal. Local opposition centers on noise and heat, not labor competition.
By 2035, clusters where power plants, substations, fiber, and a few hundred workers define the “city” become plausible, machine-first zones where human settlement is incidental.
Heat reuse adds revenue
MintGreen in British Columbia pipes immersion-cooled mining heat into a municipal district-heating network, claiming it can displace natural gas boilers. Norway’s Kryptovault redirects mining heat to dry logs and seaweed.
MARA ran a pilot in Finland where a 2 MW mining installation inside a heating plant provides a high-temperature source that would otherwise require biomass or gas.
A miner paying rock-bottom power rates can also sell waste heat, running two revenue streams from the same energy input. That makes cold-climate sites with district-heating demand newly attractive.
Kentucky’s HB 230 exempts electricity used in commercial crypto-mining from state sales and use tax.
Supporters concede that the industry creates few jobs relative to the size of the power subsidy. Bhutan’s partnership with Bitdeer bundles sovereign hydropower, regulatory support, and a $500 million fund.
El Salvador wrapped its geothermal plan and Bitcoin City in legal tender status, tax breaks, and preferential access to geothermal energy from volcanoes.
The policy toolkit includes: tax exemptions on electricity and hardware, fast-track interconnection, long-term PPAs for curtailed power, and, in some cases, sovereign balance sheets or legal-tender experiments.
Jurisdictions compete to deliver the cheapest, most reliable stream of electrons with the fewest permitting hurdles.
What’s at stake
For two centuries, industrial geography optimized for moving raw materials and finished goods through ports and railheads, with cheap labor and market access as co-drivers.
The Bitcoin mining boom is the first time we’ve had a global, capital-intensive industry whose product is natively digital and whose primary constraint is energy price.
That has revealed where the world’s “wasted watts” live and how much governments are willing to pay, in tax breaks, interconnection priority, and political capital, to turn those watts into hash.
If AI and generic compute adopt the same mobility, the map of future data centers will be drawn less by where cheap hands live and more by where stranded electrons, cool water, and quiet permitting coexist. Transmission buildouts could erase the curtailment edge.
Policy reversals could strand billions in capex. AI’s latency requirements may limit the amount of workload that can be migrated. And commodity cycles could collapse hashrate economics entirely.
But the directionality is visible. Bhutan monetizes hydro through hash. Texas pays miners to shut off during heat waves.
Kentucky exempts mining electricity from tax. China’s miners quietly reboot in provinces with surplus power. These are jurisdictions rewriting the bidding rules for compute-intensive industry.
If the industrial age organized around hands by the harbor, the compute age may organize around watts at the edge. Bitcoin is just the first mover exposing where the map already wants to tear.