For more than a decade, Bitcoin mining ran on a simple formula: find cheap electricity, run efficient machines, and mine enough Bitcoin to cover costs with a margin left over. Low power prices decided where miners built and whether they made money.
That formula is breaking down. AI data centers, new manufacturing and electrification are competing for the same constrained grid capacity miners spent years securing. The EIA expects U.S. electricity use to set records in both 2026 and 2027, rising from 4,195 billion kWh in 2025 to about 4,356 billion kWh in 2027. In that market, cheap power alone isn't a durable advantage. As CryptoSlate put it, miners now have to show they deserve the power they hold.
I covered the reliability side of that test in Bitcoin Miners Have Until 2027 to Prove They Belong on an Overloaded Grid. This article looks at the business side: how the economics of a mining site change when its most valuable asset may no longer be its hashrate, but its megawatts.
The megawatt is now the asset
A mining campus that made sense because of its cheap power may now be sitting on infrastructure other customers would pay far more to use. Land, a substation, an interconnection agreement and hundreds of megawatts of energized capacity are exactly what AI developers are struggling to find, and what can take a new project years to secure.
That changes how a mining site should be valued. The old question was whether the site could mine profitably at its contracted power price. The new question is what the highest-value use of the site's power is, hour by hour and year by year. Bitcoin production is one answer. It's no longer the only one.
Curtailment is a second revenue line
A flexible miner earns money two ways: from the Bitcoin it mines, and from the power it chooses not to use.
Start with what mining actually earns per unit of electricity. According to MillionMiner, hashprice was about $40 per petahash per day in early October 2026. A modern fleet running at about 20 joules per terahash uses roughly 0.48 MWh a day for each petahash, which works out to about $83 of mining revenue for every MWh consumed, before pool fees and other costs.
That number is the miner's break-even power price. When wholesale power costs $40/MWh, mining makes sense. During a heat wave like this summer's in Virginia, when wholesale prices jumped from about $40 to more than $600/MWh, the math flips. For a 100 MW site, an hour of mining would cost about $60,000 in electricity to earn about $8,300 in Bitcoin. A site exposed to spot prices that curtails for that hour comes out roughly $51,700 ahead.
The break-even also moves. If Bitcoin's price doubles and network difficulty holds steady, hashprice roughly doubles, and the break-even power price roughly doubles too.
Mining revenue per MWh at different hashprices
WattThe?! estimate. Assumes a fleet efficiency of 20 J/TH; excludes pool fees and other operating costs.
That's both the opportunity and the weakness. Low hashprice makes miners very willing to curtail. High hashprice makes them much less willing, even when the grid needs the relief most.
Five ways a site can earn from its megawatts
Depending on the market, the value of flexibility can show up in several places:
| Revenue line | How it works | Tied to Bitcoin's price? |
|---|---|---|
| Bitcoin mining | Hashrate earns block rewards and fees | Yes, fully |
| Curtailment and demand response | Avoided power purchases or payments for cutting load on request | Partly, since the trade-off shifts with hashprice |
| Peak-charge avoidance (Texas 4CP) | Curtailing during the four summer peak intervals cuts next year's transmission charges | No, though the cost of curtailing is |
| Battery services | Energy arbitrage, ancillary services and ride-through support | No |
| AI and HPC hosting | Long-term leases of power, space and cooling | No |
The peak-charge line alone is significant. In Texas, coincident-peak transmission charges are set by a site's demand during the four highest 15-minute intervals of each summer. Analysts cited by TheEnergyMag estimate full exposure at about $50,000 per MW per year, roughly $5 million a year for a 100 MW site. A miner that reliably steps aside during those intervals keeps most of that money.
Notice the pattern in the table: the further down the list, the less the revenue depends on Bitcoin. That's the direction the industry is moving.
Why grid operators want contracts, not goodwill
A 2026 working paper on Texas mining loads, cited by CryptoSlate, confirmed what the math suggests: miners respond to wholesale prices and peak charges, but that response weakens as hashprice rises. A grid operator can't plan reliability around a load whose flexibility shrinks whenever Bitcoin rallies.
Texas has started closing that gap. Under Senate Bill 6, large loads connected at transmission level after December 31, 2025 must be curtailable during grid emergencies, and ERCOT can direct sites with backup generation to run it or reduce load. Regulators are also developing a reliability service that would pay loads able to shed on command.
For miners, that's a trade. Mandatory curtailment removes some choice. A paid, contracted service turns flexibility into a predictable revenue stream that doesn't depend on Bitcoin's price. The miners that sign up early and perform well will be the ones regulators trust.
Flexible load isn't the same as grid-friendly load
A site that switches off when prices spike isn't automatically good for the grid. According to Texas compliance analysis, ERCOT has recorded 28 events since January 2023 in which large computing loads dropped at least 100 MW during grid disturbances. Controlled curtailment helps. Thousands of megawatts disappearing during a voltage dip does the opposite.
ERCOT's NOGRR 282 rule, effective August 1, 2026, now requires new large computing loads of 75 MW or more, crypto mining included, to ride through defined disturbances. That's an engineering problem, not a market one: protection settings, power electronics and controls. Industry estimates put compliance at $500,000 to $1 million per MW for some sites, a real cost that needs to be in every project model.
Renewable surplus is the best-case use
The strongest version of the miner argument sits next to wind and solar. Renewables produce when the weather allows, not when customers need them, and in congested areas that surplus is often curtailed or sold near zero. A mining load that ramps up in those hours and down when the grid tightens turns wasted energy into revenue and improves the renewable project's economics.
Being located near a wind farm proves nothing on its own, though. What matters is the consumption profile: whether the site runs in surplus hours, responds to congestion and stays off during scarcity. Automated controls tied to prices, generation forecasts and grid signals are what turn that claim into documented results.
Batteries and miners are a natural pairing
Mining fleets and battery storage offer different kinds of flexibility. Miners can cut load. Batteries can cut load, discharge power and respond in milliseconds.
Paired together, a battery handles the fast jobs: smoothing ramps when thousands of machines switch on or off, supporting ride-through during voltage events and covering the first minutes of a curtailment call. The mining fleet handles longer curtailment. Depending on market rules, the battery also earns its own revenue through energy arbitrage and ancillary services, independent of Bitcoin's price.
It isn't automatic. Batteries add capital cost, degradation and interconnection complexity, and the right size depends on what the local grid needs and pays for. You can test the sizing with the Utility-Scale BESS Sizing Calculator.
The AI pivot: mining Bitcoin or monetizing power?
For the largest miners, the question is whether mining is still the best use of their power at all. According to MillionMiner, public miners moved about 75 EH/s of capacity away from Bitcoin in the first half of 2026. IREN plans to exit Bitcoin mining by the end of 2026 and Cipher by the end of 2027. Hut 8 doubled its Beacon Point campus in Texas to 704 MW under AI contracts. Core Scientific now earns 83% of its revenue from colocation, and TeraWulf 71% from HPC leases.
AI tenants offer what mining can't: long-term contracts with predictable revenue that doesn't depend on Bitcoin's price, network difficulty or the next halving.
The conversion isn't simple. AI facilities need much higher power density, liquid or advanced cooling, electrical redundancy, low-latency fiber and high-uptime systems. An energized mining site gives a developer a head start, but it's closer to building a new data center on an existing power connection than to a retrofit.
The likely result is a split industry. Some operators will become diversified power-and-computing companies that lease part of their capacity to AI tenants and mine with the rest. Others will stay pure-play miners where flexible, low-cost power still works. Either way, it reinforces the point from Is Compute the New Oil?: the commodity underneath is electricity.
What 2027 does and doesn't mean
2027 isn't a federal deadline, and no single rule decides every miner's future. It's a proving period. As I explained in the 2027 article, utilities and grid operators will soon have much more data on how large flexible loads behave during scarcity, voltage disturbances and renewable surpluses. Operators with documented performance will negotiate from strength.
What this means for developers and operators
Before committing capital to a flexible load, I'd want answers to a few questions:
- At what hashprice does curtailment beat mining for this fleet, and how often do local prices clear above that level?
- What is the site's peak-charge exposure, and how reliably can it avoid those intervals?
- What will ride-through compliance cost, and could a battery cover it while earning revenue?
- Is there a paid demand response or reliability program, and what does it require?
- Would an AI tenant pay more for the same megawatts, and what would the conversion cost?
The bottom line: value per megawatt is the new metric
Bitcoin mining isn't automatically a grid benefit because its machines can be switched off, and it isn't automatically a bad load because it uses a lot of power. What matters is whether a specific site delivers measurable value where and when it operates.
The next advantage in mining may not go to the operator with the cheapest power. It will go to the one that gets the most value from every megawatt it controls, through Bitcoin, demand response, renewable integration, battery services, AI leasing or a mix of all of them. Utilities are no longer asking how much power a miner can get. They're asking what the miner offers in return.
Chris Kalowes
Founder, WattThe?!
Energy intelligence. Simplified.
