A September debate over whether Bitcoin mining electricity use has already peaked now turns on a narrower data question: Cambridge’s published index does not place the 2025 high in December at roughly 190 TWh.
In a September 15 essay, economist Saifedean Ammous argued that Bitcoin mining electricity consumption and capital expenditure may have entered a long-term decline as halvings compress subsidy growth and AI raises the opportunity cost of power and data-center infrastructure.
The broader thesis is explicitly presented as a testable hypothesis. The disputed part is the supporting date: Ammous cited roughly 190 TWh annualized in December 2025 as a possible peak, but a later review of the Cambridge Bitcoin Electricity Consumption Index points to a higher published reading earlier in the year.
Cambridge’s published 2025 high came earlier
A September 19 claim check reviewed Cambridge’s published daily best-guess series and found a maximum annualized estimate of 227.43 TWh on October 18, 2025. It reports that December 2025 ranged from 189.56 to 211.75 TWh, meaning 190 TWh was near the bottom of that month’s range rather than the index maximum.
That correction matters because “Bitcoin electricity use peaked” and “Cambridge’s index peaked in December at 190 TWh” are different claims. The second can be checked directly against the published series. The first is a longer-term hypothesis that only future data can confirm or falsify.
Cambridge itself describes CBECI as a model-based estimate of network power demand rather than a direct meter reading of every mining site. The index applies assumptions about hashrate, hardware efficiency, electricity prices and miner profitability, then reports a best-guess annualized estimate.
The broader hypothesis survives the date correction
Correcting the 2025 peak date does not by itself settle the larger argument. Ammous’s thesis is that subsidy growth eventually becomes too weak to support ever-rising mining capital expenditure and electricity demand, especially when miners face competing uses for the same power infrastructure.
That opportunity-cost argument fits a trend BitcoinVersus.tech has already documented. Mining companies increasingly value energized land, substations and grid interconnections as AI infrastructure, sometimes more than the ASICs that originally justified those sites.
If an AI tenant can pay more for the same megawatt, a mining operator has an economic reason to redirect capital toward compute infrastructure. That does not require Bitcoin mining to disappear. It only means mining must compete with another buyer for scarce power, buildings, transformers and fiber.
Idle ASIC capacity shows electricity demand can move quickly
Bitcoin mining power demand is also more flexible than a simple installed-capacity chart suggests. BitcoinVersus.tech recently reported that roughly 235 EH/s of ASIC capacity may be sitting idle, highlighting how quickly changes in profitability, hosting conditions or electricity prices can remove machines from production.
That flexibility cuts both ways. A falling annualized electricity estimate may reflect machines shutting down, but those machines can return if hashprice, Bitcoin price, power cost or network conditions improve. A temporary decline therefore does not automatically establish a permanent peak.
Efficiency can rise even if total power falls
There is another reason the network can keep getting computationally stronger without consuming ever more electricity: ASIC efficiency keeps improving. BitcoinVersus.tech recently examined how lower joules per terahash allow more hashes from the same power envelope.
If older machines retire and newer machines replace them, total hashrate can recover or rise even while electricity demand stays flat or falls. That is the scenario behind Ammous’s distinction between a possible peak in mining energy expenditure and a network that can still become more computationally efficient.
What would actually prove a peak?
A durable conclusion requires more than one down-cycle. The strongest evidence would be a long period in which comparable Cambridge best-guess estimates remain below the October 2025 high even through stronger Bitcoin prices, higher transaction fees and improved miner economics.
Conversely, a sustained move above Cambridge’s published 227.43 TWh annualized high would weaken the claim that the network’s electricity demand had already topped out in 2025.
For now, the cleanest conclusion is narrower: Bitcoin mining electricity estimates have fallen materially from their 2025 highs, but the permanent peak remains unproven. The debate is now less about whether power demand fell and more about whether the economic forces behind that decline are cyclical or structural.
BitcoinVersus.Tech
Advertisement
BitcoinVersus.Tech Editor’s Note:
We volunteer daily to ensure the credibility of the information on this platform is Verifiably True. If you would like to support to help further secure the integrity of our research initiatives, please donate here: 3C9o19EH5HSiwEPyCTmEKzxhNCbo2X6TTb
BitcoinVersus.tech is not a financial advisor. This media platform reports on financial subjects purely for informational purposes.
Leave a comment