Irish Study Models Bitcoin Mining as Supply-Side Flexibility for Wind Farms

Researchers at the Technological University of the Shannon have published a study that models Bitcoin mining as a flexible electricity consumer for Irish wind farms. The paper, "Bitcoin mining as supply-side flexibility in Irish wind energy integration" by Marcel Sarnecki and Niall Burke, was submitted in November 2025 and appeared in the journal Energy Economics in August 2026.

Using hourly 2024 data from the Irish electricity market, the authors simulate a 100 megawatt wind farm coupled directly with an on-site mining operation. Their central result is that mining can consume a large share of wind energy that would otherwise go unused, and that under specific operating models it also raises the wind farm's total revenue.

A growing amount of wind is curtailed

Ireland switches off a rising share of its wind generation, a practice the grid calls dispatch-down. In 2014 the country had roughly 2.5 gigawatts of installed wind capacity and curtailed about 4 percent of available wind energy. By 2024 installed capacity had climbed to about 4.9 gigawatts, and curtailment had reached 10.1 percent, or 1.3 terawatt-hours. The share rose again to 11.4 percent in 2025. The study notes that curtailment is heavily regional, averaging about 19 percent in the northwest against roughly 9 percent in the midlands, because Ireland's wind sits on the Atlantic coast while its demand sits in the east.

Ireland aims to reach 8 gigawatts of onshore and 5 gigawatts of offshore wind by 2030. The study notes that transmission networks are not expanding as fast as generation, which widens the gap between available wind power and what the system can absorb.

What the study modeled

The authors run the wind farm across all 8,760 hours of 2024, drawing on public data for wind speed and output, Irish Single Electricity Market prices, curtailment, and the Bitcoin price and network hashrate. For each hour the model allocates the available energy to maximize the system's total revenue.

They compare six configurations. A no-mining baseline (S0) sells wind to the grid or curtails it. S1 mines only with energy that would otherwise be curtailed. S2 adds regular wind power when mining earns more than selling it. S3 adds grid electricity when mining covers the cost including network charges. S4 allows either own wind or grid power after curtailed energy, and S5 allows both at once. Mining capacity is varied from 5 to 90 megawatts in 5 megawatt steps.

For hardware, the authors compare an older Antminer S9 at about 98 joules per terahash with a newer S21 Hydro at 16 joules per terahash. They base the analysis on the S21 Hydro, because the far less efficient S9 was not economic under the modeled conditions. The paper studies newly built commercial rigs with full investment costs, and does not test already-owned or second-hand hardware run on free power.

A 20 to 30 megawatt rig absorbs most curtailed power

The modeled farm could have produced 288,831 megawatt-hours in 2024, of which 33,808 megawatt-hours were curtailed. Because the model always uses curtailed energy first, the share absorbed depends on the size of the mining rig rather than on the scenario.

A 20 megawatt rig captures 83.1 percent of the curtailed energy, a 25 megawatt rig about 90 percent, and a 30 megawatt rig roughly 93 percent. Beyond about 60 megawatts the operation absorbs almost all curtailed wind, and further capacity adds little.

Curtailed power alone does not pay

Free power does not mean free mining. The study finds that curtailed energy on its own does not run a newly built rig often enough to justify its cost. A 20 megawatt operation would absorb 83 percent of curtailed energy but sit at only about 16 percent utilization, against roughly 34.7 million Euro in hardware and infrastructure cost and about 420,000 Euro in annual fixed operating cost.

Utilization falls as the rig grows. In S1, which mines only curtailed power, a 5 megawatt rig runs at about 30 percent, a 20 megawatt rig at about 16 percent, and a 90 megawatt rig at under 5 percent. The authors call S1 not financially viable for a new commercial operation. The limiting factor is utilization against capital cost, not the price of power.

Flexible models raise revenue

The models that add other energy sources reach much higher utilization, from 31 to 76 percent. A 20 megawatt rig in S2 runs at a 61 percent mining capacity factor and a 30 megawatt rig at 52 percent, far above the roughly 16 percent that curtailed power alone would sustain. In the baseline, selling wind at the market earned 22.2 million Euro, an average of 86.90 Euro per megawatt-hour.

In S2, a 20 megawatt rig that also mines saleable wind when that earns more than selling it raises total revenue by 32 percent, to 29.2 million Euro, and lifts the average to 103.10 Euro per megawatt-hour. At 30 megawatts revenue rises 40 percent to 31.1 million Euro. The wind farm's effective capacity factor improves from 29 to 32 percent. The shortest payback periods, about 2.2 to 4.4 years under 2024 conditions, occur at 5 to 20 megawatts in the flexible scenarios.

Returns hinge on Bitcoin outpacing hashrate

The revenue gains rest on the Bitcoin price and the network hashrate. The study's central point here is that viability depends on the spread between the two, not on the price level alone. When price and hashrate grow at the same rate, mining revenue per megawatt-hour stays broadly flat and returns come from absorbing curtailment rather than from price appreciation. Positive returns over the six-year horizon require the price to grow faster than hashrate.

The absolute price still sets a floor. In the study's static payback table for a 20 megawatt operation, the investment pays back within six years at a Bitcoin price of 100,000 Euro across all curtailment levels, but at 80,000 Euro only the most curtailed sites do, and at 60,000 Euro none do. The model used a base price near 90,000 Euro and a network hashrate of 780 exahashes per second. Hashrate has since run near 900 exahashes per second, and in late August 2026 Bitcoin traded near 79,000 Dollar, roughly 73,000 Euro.

The authors call their revenue figures conservative in one respect, because they exclude grid-service income such as fast frequency response that a 20 megawatt load could also earn. They also note the model assumes perfect foresight of prices, which pushes the revenue estimates toward an upper bound.

The authors' framing

Sarnecki and Burke treat mining as a privately financed, market-driven investment rather than a subsidized one, and stress that it cannot replace the grid expansion a high-renewables system needs. They present it as one additional way for wind farms with high curtailment to monetize otherwise unused energy, which could in turn make grid investment easier to finance. Whether it pays, they write, depends on rig sizing, hardware efficiency, the regulatory setting, local market conditions, and the future path of the Bitcoin price and network hashrate.

For why the profitable configurations look different from the curtailment-only one, see our analysis on what mining actually does for a wind farm.