Bitcoin's power consumption may have peaked, with December 2025 potentially marking its all-time high.

Source: Saifedean Ammous, author of *Bitcoin Standard*; Translated by Jinse Finance Claw

For several years, I've had a theory about why and how Bitcoin mining has stopped growing in terms of capital expenditure and electricity consumption. I originally expected this dynamic to unfold gradually over many years, so I never wrote it down clearly; but some interesting developments over the past year make me feel that this shift may be happening now.

Bitcoin's electricity consumption may have peaked, and it may never consume as much electricity again as it did in 2024-2025. Even if this isn't the case, I expect this dynamic to become a reality sometime in the coming years due to mining economics and the halving mechanism. This article presents this argument as a testable hypothesis, provides data to test it, and discusses its impact on the Bitcoin mining industry and Bitcoin itself. The assumption is that the Bitcoin halving means that Bitcoin mining will continue to grow as long as the expected increase in Bitcoin's market capitalization exceeds 100% over four years (i.e., 18.92% per year). Considering the long-term depreciation of the US dollar, this figure would be several percentage points higher when measured in US dollars. The opportunity cost of Bitcoin mining is rising as AI has become a major source of demand for electricity and data center infrastructure, further increasing this figure by several percentage points. Examining Bitcoin's price performance and mining metrics reveals that this threshold has been crossed, and Bitcoin mining has shifted from long-term growth to long-term decline this year. Because the number of newly mined Bitcoins halves every four years, the market value of mining output increases if the price of Bitcoin rises by more than 18.92% annually plus the depreciation of the US dollar; if the increase is less than this level, it decreases. When miners—and Bitcoin holders more broadly—expect an annual price increase of Bitcoin higher than 18.92% plus the depreciation, they are marginally more likely to increase their mining investment, leading to increased capital expenditures and electricity consumption. When they expect an annual price increase of less than 18.92% plus the depreciation, they are marginally more likely to reduce their mining investment, causing the Bitcoin mining industry to shrink in terms of capital expenditures and electricity consumption. Inevitably, there will come a point where the market capitalization of Bitcoin no longer appreciates at a rate exceeding 100% every four years. Early on, Bitcoin was an asset with a very small market capitalization. As more people learned about it, it grew rapidly from a very low base, achieving a high annual growth rate. As Bitcoin's price rose, its market capitalization also grew, and maintaining the same growth rate would require increasingly larger inflows of capital. As Bitcoin absorbs a larger share of global currency demand, maintaining any fixed percentage appreciation would require holders to collectively hold an ever-increasing absolute amount of real balance denominated in Bitcoin from a limited global stock of savings and wealth. As expected, Bitcoin's price growth rate has been slowing over time. And if this growth continues indefinitely, it will eventually swallow up all other forms of currency, reaching a point where it becomes the world's only currency whose market value can only grow at the rate of growth in savings and the output of goods and services, likely only a low single-digit percentage. Between now and then, Bitcoin's price appreciation will reach a tipping point, and mining will begin to contract. Bitcoin Price Expectations Predicting Bitcoin's future price is difficult, but the decline in its price growth rate is quite predictable. Once the annual growth rate—and more importantly, market participants' expectations of the growth rate—practically falls below 18.92%, I would expect Bitcoin mining to decline in both the capital it attracts and the energy it consumes. Bitcoin is extremely volatile, so estimates of the current price growth rate depend heavily on the chosen time window. Over the past year, it has fallen by 29%. Over the past two years, it has risen by an average of 16% annually. Over the past three years, the compound annual growth rate was 45%; over the past four years, it was 42%; and over the past five years, it was only 10%. The five-year period is so low because this particular five-year period begins near the top of the last bull market and ends today, likely near the bottom of the current bear market. But if we extend the timeframe to 6, 7, and 8 years, the compound annual growth rates are 38%, 35%, and 35%, respectively. Investors seeing these figures would likely assume that Bitcoin's price will rise by roughly 5% to 40% annually over the next few years. Five years ago, the calculated compound annual growth rate would have been much higher, and correspondingly, the expected return would have been much higher. The fact that the current five-year return is only around 10% is likely discouraging people from investing in Bitcoin mining. A miner who finds the price only rising by 10% annually could suffer significant losses even in fiat currency terms, let alone compared to directly holding Bitcoin. This is likely to reduce investment in Bitcoin mining. Mining Rewards for Each Mining Cycle We can also understand this assumption by examining the evolution of the average mining reward across different mining cycles (defined as every 210,000 blocks, approximately four years, during which the Bitcoin mining subsidy remains constant). The table below lists the average daily Bitcoin price for all trading days within each mining cycle, along with the multiple of the price for each cycle relative to the previous cycle. The last column is the multiple of the block reward in USD relative to the block reward in the previous cycle, which is exactly half the price multiple because daily mining is halved each cycle. As can be seen from the table, the dollar block reward revenue in the second cycle was more than 30 times that of the first cycle, the block reward in the third cycle was 8 times that of the second cycle, and the fourth cycle was 2.86 times that of the third cycle. As expected, during these periods, Bitcoin's difficulty, hashrate, and electricity consumption all increased very rapidly, but the rate of increase decreased over time. The most interesting phenomenon here is that, so far, the mining block reward in the fifth cycle is only 28% higher than the previous cycle. This assumption can also be stated as: Bitcoin mining will eventually stop growing when the block reward multiple is less than "1 plus the dollar's depreciation over four years". The table below lists the minimum multiples required to incentivize Bitcoin mining investment growth under different dollar depreciation rates. The required compound annual growth rate for Bitcoin is calculated by compounding the dollar's depreciation over four years with an annual threshold return of 18.92%. The required periodic block reward multiple is half the resulting price multiple. Based on investors' estimates of the annual depreciation of the US dollar, the last column tells us how large a cycle block reward multiple is needed to incentivize investment and drive Bitcoin mining growth. Comparing the last two columns of the previous two tables, we can see why Bitcoin mining may no longer be growing. The current cycle has lasted 28 months, and block reward revenue is only 28% higher than the previous cycle. This means that for anyone who believes the dollar depreciates by more than 7% annually, the subsidies haven't kept pace with the dollar's depreciation; and even at a 6% depreciation rate, they've only barely kept pace. Given this decline in mining rewards, a slowdown or even contraction in Bitcoin mining is to be expected. Unless there's a significant reversal in this metric, this trend could continue indefinitely. This pattern is also reflected in the significant decrease in Bitcoin mining difficulty over the past ten months. Bitcoin's mining difficulty reached an all-time high on October 30, 2025, and 305 days later, it has yet to recover or surpass that level. This is the second longest period during which Bitcoin's difficulty has failed to reach a new all-time high. The longest such period was from August 1, 2011, to July 30, 2012, taking 363 days to reach a new all-time high. That decline was accompanied by the largest price correction in Bitcoin's history, with the price falling from $30 in June 2011 to $2 in November 2011, a 94% pullback. The third longest period of difficulty began on May 16, 2021, and lasted 252 days until January 23, 2022, when the difficulty reached a new all-time high. That round occurred after China banned mining, a significant event as China at the time accounted for a large share of Bitcoin mining, forcing many miners to liquidate their operations, sell their machines, or move them overseas. The need for liquidation likely also meant selling some of their Bitcoin holdings, triggering a sharp drop in Bitcoin's price. What's special about this current difficulty decline is that, unlike the first and third longest periods, it didn't occur after a dramatic event. Bitcoin's maximum retracement from its October 2025 all-time high is currently only 54%, far lower than the 95% in 2011; nor has it been accompanied by a major regulatory event affecting Bitcoin miners like in 2021. However, the difficulty has already dropped significantly, and if it takes another two months to recover, this will be the longest period of difficulty decline in Bitcoin history. March 14, 2024: Peak Bitcoin Mining Rewards Another noteworthy fact is that, with two years and four months having passed in this fifth Bitcoin cycle, the mining reward per block is 3.125 bitcoins, and the daily mining reward in US dollars has not yet exceeded the highest level set in the previous cycle—that was on March 14, 2024, when the price was $73,800 and the reward per block was 6.25 bitcoins. On that day, assuming 144 blocks were produced, the total mining reward was equivalent to $65 million ($144 blocks × $73,800 per bitcoin × 6.25 bitcoins per block). March 14th was the highest reward of the previous cycle, and for the mining reward to match this cycle after the halving, the price needs to reach $147,600. Since the highest price this cycle only reached $126,000, we haven't seen a day with a higher dollar-denominated block reward value this cycle. If the price fails to reach $147,600 before the next halving, this will be the first cycle in which the daily mining reward fails to exceed the previous cycle's high. March 14, 2024, may even go down in history as the day with the highest Bitcoin reward value ever. With each upcoming halving, the price needs to double again to match the market capitalization rewarded on that day. If the Bitcoin price doesn't reach $147,600 in the fifth cycle before the next halving, then to surpass the value of March 14, 2024, in the sixth cycle (2028-2032), the price would need to reach $295,200; in the seventh cycle (2032-2036), it would need to reach $590,400; and in the eighth cycle, it would need to reach $1,180,800. The required price will continue to double every four years. I'm not suggesting that any of these values ​​are unattainable; I'm simply saying that they might not be achievable. And if they are, March 14, 2024, will forever remain the most valuable day in Bitcoin mining history. More specifically, even if one of these cycles surpasses the value of March 14, 2024, there will inevitably come another day when its mining reward value will never be surpassed, because Bitcoin's value cannot double every four years indefinitely. It's hard to imagine its market capitalization doubling every four years once the whole world uses Bitcoin as currency. If it's not March 14, 2024, then some other future date will mark the peak of Bitcoin mining. Whether the peak of Bitcoin mining has already occurred can only be answered by time; but a stronger argument is that such a peak will inevitably occur at some point. By then, Bitcoin mining will have peaked in both capital investment and electricity consumption, but mining difficulty will recover and likely continue to rise as mining equipment efficiency improves. Peak Electricity Consumption The most direct test of this hypothesis comes from Cambridge University's estimates of Bitcoin's electricity consumption. Preliminary data released by the Cambridge Centre for Alternative Finance (CCAL) indicates that the annualized electricity demand for Bitcoin mining in December 2025 was approximately 190 terawatt-hours (TWh). Subsequently, the Cambridge Bitcoin Electricity Consumption Index (CBEE) declined significantly, with the mid-2026 estimate falling back to the high end of the 130 TWh range. This provides a clear empirical test for the hypothesis. The annualized level of approximately 190 TWh in December 2025 may prove to be the peak of Bitcoin's electricity consumption rate. If the best-in-class Cambridge estimate subsequently continues to rise above this level, then this specific assertion will be disproven. However, the decline since December is consistent with the broader argument discussed above: reward revenue growth is slowing, difficulty has been unusually low for an extended period, and the amount of electricity that miners can profitably consume is beginning to shrink.

AI Accelerator

Another reason to believe we may have witnessed the peak of Bitcoin mining is the rise of AI as a major source of data center and electricity demand. Over the past four years, the number of data centers has surged to meet the explosive demand for AI, providing Bitcoin miners with another lucrative revenue stream. Rumors and observations of major Bitcoin mining companies suggest a massive shift in mining infrastructure towards AI data centers.

This indicates that Bitcoin mining has entered a new era: it is no longer the only way to monetize cheap and discarded energy, as AI can now also compete for these resources; and the massive investments in AI may be accelerating the decline in Bitcoin mining capital expenditures and electricity consumption. After years of investment in Bitcoin mining infrastructure, much of the discarded energy used for Bitcoin mining is now more sophisticated and better connected to the internet, making it more suitable for AI data centers.

The key difference between AI data centers and Bitcoin mining is that Bitcoin's difficulty adjustment makes it a zero-sum game, while AI data centers are a positive-sum game. As AI increases user productivity, they will continue to invest more in it, demanding increasingly larger workloads. However, Bitcoin mining is competing for a shrinking Bitcoin pie, halving in size every four years. Difficulty adjustments further ensure that only the most efficient miners can profit, and as more miners join and rewards halve without a corresponding doubling of revenue, profit margins will only become thinner. AI, on the other hand, could bring huge productivity gains across countless industries, thus supporting ever-increasing demand. Bitcoin miners now face a dilemma: declining expected returns from Bitcoin halvings versus potentially exponentially increasing returns from AI productivity gains. It's possible that AI will ultimately prove to be a huge bubble, with a large portion of these data center expenditures becoming waste. Chinese AI models appear capable of achieving results similar to US models with minimal energy and processing power costs. If most AI spending ultimately proves to be a wasteful misinvestment (unlikely), miners will return to Bitcoin mining, but this won't change a fundamental judgment: Bitcoin mining will peak at some point in the future. Transaction Fees To simplify the analysis, this article omits one factor: the growth of transaction fees. For most of Bitcoin's history, fees have only accounted for a small portion of miners' revenue. If fees were to rise significantly, the following analysis might become invalid, as mining rewards (including subsidies and fees) could rise to levels exceeding those of the previous cycle. However, the Bitcoin network has grown significantly, while fees denominated in Bitcoin have remained flat or even decreased over time; with the continued development of Layer 2 solutions, it's likely that fees will not rise significantly, and total mining rewards will never exceed the levels of the previous cycle. In my view, the idea that Bitcoin transaction fees need to reach a certain "security budget" is largely misleading. I don't believe fees need to grow to any specific level to ensure network security; a network is secure because there's the possibility of paying fees to circumvent mining censorship, not because fees have reached a certain level. The impact: If this assumption holds true, it's not a problem for Bitcoin. Bitcoin's supply continues to grow as planned, and as more people understand this, and understand that the supply of all other monetary assets is growing much faster, demand for Bitcoin is likely to continue to increase. We will enter a new era: Bitcoin is growing, while the scale of Bitcoin mining is shrinking. The impact on Bitcoin mining is that its scale is likely to shrink over time, and its profitability may also decline. This could very well lead more Bitcoin holders to shift their investments from mining to directly buying Bitcoin, which is likely to be bullish for the price. As Bitcoin's energy consumption decreases, the hysteria surrounding Bitcoin's energy consumption is likely to subside, as is the large amount of negative reporting. This is also bullish for the price. The actionable takeaway is: if you understand Bitcoin's value proposition, buy Bitcoin directly. If you have cheap electricity and the ability to build mining facilities, then turn to AI data centers. Conclusion: Bitcoin mining once had a structural long-term tailwind: the rate at which the monetization of block rewards increased far faster than the rate at which the number of Bitcoins was reduced by halvings. This tailwind has disappeared. Despite a 2.56-fold increase in the average Bitcoin price this cycle, nominal block reward revenue was only 28% higher than the previous cycle; after accounting for the depreciation of the US dollar, this increase is close to zero. Meanwhile, mining difficulty is experiencing its second-longest period of failure to reach new highs in history, Cambridge's estimated electricity consumption has fallen sharply from its potential peak in late 2025, and AI is increasing the opportunity cost of high-quality mining infrastructure. Bitcoin's fundamentals suggest it may continue to grow, but Bitcoin mining and its electricity consumption may have already peaked.