Bitcoin mining: the cost of scarcity
Energy consumption as the foundation of proof of work
Bitcoin has operated since 2009 without any central authority determining which transactions are valid. The system is held together by mining, the activity through which thousands of operators contribute computing power, collect transactions into new blocks, and help secure the network in exchange for newly issued bitcoin and transaction fees. Mining requires large amounts of electricity, but its impact depends not only on total consumption. The sources from which that energy is generated also matter.
Energy consumption is neither a side effect nor an accidental flaw, but the result of one of the most important and debated design choices in the entire system. Linking network security and the protection of its programmed scarcity to a real energy cost, rather than to less resource intensive mechanisms, helps bring Bitcoin closer to the economic model of gold.
The double spending problem
Any digital payment system without a central authority must answer a fundamental question: how can the same value be prevented from being transferred twice? In the traditional banking system, banks maintain the definitive record of balances and transactions. Bitcoin had to solve the problem without intermediaries by enabling thousands of independent participants, with no need to trust one another, to converge on the same payment history. The solution was proof of work, a mechanism that grants the right to write the next block to whoever demonstrates that they have incurred a real computational cost, expensive to produce but verifiable within seconds.
In practice, miners collect pending transactions, build a candidate block, and repeatedly modify a numerical value known as the nonce. With each variation, they calculate the block’s hash, a fixed length code derived from its data. Changing even a single element produces a completely different result. The objective is to obtain a hash below the threshold established by the network. There are no mathematical shortcuts. Miners must proceed through trial and error, billions of times per second, and this competition translates directly into electricity consumption. The first miner to find a valid hash broadcasts the block to the network. It is then checked by nodes, computers that run the Bitcoin software and maintain a copy of the blockchain. If the block and its transactions comply with the protocol rules, the nodes accept it and the miner receives the newly issued bitcoin, together with the fees paid by users.
Alternative mechanisms, such as the proof of stake system adopted by Ethereum in 2022, link the right to propose and validate blocks to the amount of cryptoassets committed as collateral by participants. Bitcoin instead anchors network security to external, physical costs associated with hardware, computing power, and electricity, rather than to resources that exist entirely within the system. In a proof of stake system, anyone who controls a sufficient share of the cryptoasset can, at least in theory, exert significant influence over the network. In a proof of work system, anyone seeking to attack Bitcoin must acquire sufficient hardware, computing power, and energy, incurring costs that cannot be avoided simply by accumulating the cryptoasset being targeted.
Bitcoin and digital gold
Proof of work protects the immutability of the transaction history and helps make Bitcoin’s programmed scarcity credible. New bitcoin are issued according to a predetermined schedule, with the amount halving approximately every four years until issuance ends around 2140. At the same time, changing a transaction that has already been recorded would require rebuilding the computational work accumulated in all subsequent blocks while continuing to compete against the rest of the network.
This combination has encouraged Bitcoin’s comparison with gold as a store of value. Both are scarce, require resources to produce, and generate neither cash flows nor intrinsic returns. There is, however, a fundamental difference. Bitcoin issuance is fixed and does not respond to price changes. Even a sharp increase in demand cannot accelerate the creation of new units. Gold production, by contrast, can increase when higher prices make previously unprofitable deposits economically viable, although the process remains subject to the timeframes and limitations of physical extraction.
The comparison does not imply equivalence. Gold also has nonmonetary demand from jewellery, electronics, and industrial applications, while Bitcoin derives its value from confidence in its scarcity and in the rules governing its operation. The origin of that scarcity is also different. Gold’s scarcity is determined by geology, whereas Bitcoin’s is established by a shared and verifiable software protocol. Proof of work helps protect this scarcity by linking network security to a real energy cost. Moving to a more efficient model, such as proof of stake, would not simply eliminate that cost. It would replace the principle underlying the protection of the network and its comparison with gold.
Energy sources
According to the 2025 Cambridge Digital Mining Industry Report, Bitcoin mining consumes approximately 138 terawatt hours of electricity on an annualised basis, equivalent to around 0.5 per cent of global electricity consumption. This figure conveys the scale of the activity, but does not by itself describe its environmental impact. The consequences of consuming the same amount of electricity can vary considerably depending on how that electricity is generated.
The same report estimates that 52.4 per cent of the electricity used for mining comes from sources classified as sustainable, comprising 42.6 per cent renewables and 9.8 per cent nuclear power. Hydroelectric power is the largest renewable source at 23.4 per cent, followed by wind at 15.4 per cent and solar at 3.2 per cent. Natural gas nevertheless remains the single largest energy source at 38.2 per cent, while coal and oil account for 8.9 and 0.5 per cent respectively. The result is an evolving energy mix that remains divided between low emission sources and fossil fuels.
Compared with gold extraction, digital mining has a distinct operational advantage. A mine is tied to the location of its deposit, whereas a mining farm can be established wherever electricity is abundant and inexpensive, and relocated when economic or regulatory conditions change. Since energy is one of the industry’s largest costs, this mobility can direct operators towards inexpensive hydroelectric, solar, or wind power, as well as towards productive capacity that would otherwise remain unused. The relationship is not automatic, however. In some regions, the least expensive electricity still comes from fossil fuels. Assessing the impact of mining therefore requires considering total energy consumption, the composition of the energy mix, and emissions intensity together.
The origins of proof of work
Proof of work predates Bitcoin by fifteen years and originated in efforts to combat email spam. In 1993, cryptographers Cynthia Dwork and Moni Naor published an academic paper titled “Pricing via Processing or Combatting Junk Mail,” in which they proposed discouraging the mass distribution of unwanted messages by requiring senders to solve a small mathematical problem first. The cost would be negligible for someone sending a single message but prohibitive for someone sending millions.
In 1997, Adam Back, a computer scientist associated with the cypherpunk movement and now cofounder and CEO of Blockstream, turned that idea into a practical system called Hashcash, which used hash functions rather than generic mathematical calculations. The term “proof of work” was formalised in 1999 in a paper by Markus Jakobsson and Ari Juels. In 2004, cryptographer Hal Finney, who would later become one of Nakamoto’s earliest direct interlocutors, proposed a variation called “reusable proof of work,” bringing the mechanism closer to the construction of a digital monetary system. When Nakamoto published the Bitcoin white paper in October 2008, Hashcash was explicitly cited as a reference for Bitcoin’s mining and consensus mechanism.
Bitcoin’s innovation therefore lies not in inventing proof of work, but in transforming it into the foundation of a decentralised network capable of operating without central authorities, intermediaries, or relationships of trust among participants.
Mining pool concentration
Proof of work security benefits from the distribution of computing power among a sufficiently large number of independent participants. In recent years, however, mining activity has become concentrated around a small number of large pools. These pools aggregate the computing power of thousands of operators and distribute rewards in proportion to each participant’s contribution, making returns more consistent. According to MiningPoolStats data from mid 2026, Foundry USA, AntPool, ViaBTC, and F2Pool together accounted for more than 70 per cent of the network’s computing power, with Foundry USA close to 30 per cent. These percentages should be interpreted with caution because they are estimates based on block attribution and do not necessarily indicate direct ownership of the computing power involved.
Pool concentration primarily affects the network’s resistance to transaction censorship. Based on the distribution of blocks observed during the period under consideration, approximately three pools would have been sufficient to exceed half of the network’s computing power. This provides an indication of operational concentration, but does not mean that pools directly own all the computing power attributed to them. Pool operators choose which transactions to include in blocks and could, at least in theory, exclude those originating from sanctioned addresses or considered undesirable. The risk is mitigated by the fact that the aggregated computing power belongs to individual miners, who can redirect it to another pool. The Job Declaration Protocol, one of the components of Stratum V2, also aims to allow miners to select the transactions included in blocks instead of receiving a complete block template from the pool. A tension nevertheless remains between Bitcoin’s decentralised architecture and the operational concentration of the mining industry.
The end of new issuance
Bitcoin’s issuance schedule provides for the amount of new bitcoin assigned to each block to be reduced by half approximately every four years through an event known as the halving. This process will continue until around 2140, when the final fraction of the 21 million bitcoin specified by the protocol will have been issued. Until then, mining revenue will come from two sources: the block subsidy, consisting of the new bitcoin assigned to each block, and the fees paid by users to have their transactions included. As successive halvings take place, the relative importance of the block subsidy will gradually decline, while fees will become increasingly important until they are the sole source of mining revenue.
The gradual transition from the block subsidy to transaction fees does not necessarily imply a weakening of the network. The new equilibrium will depend on demand for block space, the economic value of transactions, the price of bitcoin, and miner efficiency. Higher fees in absolute terms could compensate for the reduction in new issuance. If they do not, computing power will adjust to the revenue available, as already occurs following each halving.
No blockchain of comparable size has yet operated under a model supported almost entirely by transaction fees, which currently represent a minority share of mining revenue. The transition, however, will not occur suddenly in 2140. It is already taking place with each reduction in the block subsidy. The long term question therefore concerns the level of security that the network will be able to finance relative to the value it stores and transfers.
Conclusions
Bitcoin mining is the practical expression of a deliberate design choice: to anchor network security and the protection of its programmed scarcity to a real physical cost, rather than to a central authority or a mechanism operating entirely within the system. Energy consumption is therefore neither a side effect nor an accidental flaw, but an essential component of proof of work. The cost borne by miners makes producing new blocks expensive and any attempt to alter the blockchain equally costly, while verification remains simple and accessible to all participants.
This does not make either the scale of consumption or the source of the electricity irrelevant. Both remain essential to assessing its environmental impact. It does, however, mean recognising that mining’s demand for energy performs a specific economic and technological function. Moving to a more energy efficient model, such as proof of stake, would not simply optimise Bitcoin. It would change its security model and break the link between network protection, programmed scarcity, and the use of physical resources on which the comparison with gold also rests.
The strength of this model will increasingly depend on mining’s ability to remain distributed and adapt to the gradual decline in new issuance. The concentration of computing power among a few large pools and the growing importance of transaction fees are the principal tests of its evolution. They are not necessarily signs of an imminent limit, but structural issues that will shape Bitcoin’s ability to preserve security, neutrality, and decentralisation over time.
