TJE Ventures

How Proof-of-Work Mining Actually Works

Bitcoin and several other cryptocurrencies use a consensus mechanism called Proof of Work (PoW) to validate transactions and add new blocks to the blockchain. Understanding the mechanism is necessary to understand why the energy cost is not incidental but structural — and why it cannot be engineered away without changing the fundamental design of the network.

To add a new block to the Bitcoin blockchain, a miner must find a specific number — called a nonce — that, when combined with the block’s transaction data and run through a cryptographic hash function (SHA-256), produces an output below a target value. There is no shortcut to finding this number; the only method is brute-force trial: generate a candidate nonce, compute the hash, check the result, discard and repeat. The miner who finds a valid nonce first wins the block reward — currently 3.125 BTC after the April 2024 halving — and their block is added to the chain.

The network automatically adjusts the difficulty of the target every 2,016 blocks (approximately every two weeks) to maintain an average block time of 10 minutes regardless of how much mining capacity is online. If more miners join the network, difficulty rises. If miners leave, difficulty falls. This self-adjusting mechanism is the root of the energy problem: adding more efficient hardware to the network does not reduce total energy consumption — it raises the difficulty level until energy consumption returns to the economic equilibrium where the cost of mining approximately equals the value of the block reward. Efficiency improvements are competed away, not accumulated.

Modern Bitcoin mining uses purpose-built Application-Specific Integrated Circuits (ASICs) — chips designed to do nothing except compute SHA-256 hashes as fast as possible. A current-generation Antminer S21 Pro computes approximately 234 trillion hashes per second (234 TH/s) while consuming 3,510 watts. A warehouse containing 10,000 of these machines draws approximately 35 MW of power and generates 35 MW of waste heat — every hour, every day, whether Bitcoin’s price is $100,000 or $10,000.

The Scale of the Energy Consumption

Bitcoin network electricity consumption estimates, 2024 (Cambridge Centre for Alternative Finance):

• Bitcoin alone: ~120–150 TWh/year
• All proof-of-work cryptocurrency combined: ~150–180 TWh/year
• Comparable national equivalents: Argentina (~130 TWh), Poland (~150 TWh), the Netherlands (~120 TWh)

For context established in the data center article, the entire global productive data center sector consumes approximately 200–250 TWh per year. Crypto mining consumes roughly 60–70% as much electricity as all of the world’s productive computing infrastructure combined, to perform work that has no utility except proving that electricity was spent.

The geographic distribution of Bitcoin mining hashrate (2024 estimates):

• United States: ~35–40% (led by Texas, Kentucky, Georgia, and New York)
• China: ~15–20% (reduced from ~65% before the 2021 ban, but recovered via underground operations)
• Kazakhstan: ~10–15% (largely coal-powered)
• Russia: ~8–10%
• Canada: ~6–8%
• Other: ~15–20%

The carbon intensity of this consumption varies enormously by geography. U.S. mining operates on a grid mix that is roughly 40% non-emitting. Kazakhstan mining operates primarily on Soviet-era coal plants. The weighted average carbon intensity of the Bitcoin network’s electricity consumption produces an estimated 22–30 million metric tons of CO₂ per year — roughly the annual emissions of a mid-sized country like Portugal or Sweden.

The Good: Where Mining Provides Genuine Value

Stranded and Curtailed Energy

The most legitimate environmental argument for crypto mining is that it can monetize energy that would otherwise be wasted — stranded generation with no transmission access, curtailed renewables that the grid cannot absorb, or remote power sources too far from load centers to be economically delivered.

The mechanism is straightforward: Bitcoin mining requires electricity and an internet connection. It does not require proximity to customers, grid infrastructure, or any physical supply chain. A mining operation can be containerized, shipped to a remote location, and powered by whatever electricity source is available there. This portability makes miners the only large flexible load that can reach energy sources that no other industrial process can access economically.

Real-world examples where this argument holds:

• Remote hydroelectric in Iceland, Paraguay, and Bhutan. These countries generate more hydroelectric power than their domestic economies can consume. Mining operations absorb the surplus that cannot be exported due to limited interconnection capacity. The alternative to mining this energy is not using it for something more productive — the alternative is spilling water over the dam without generating electricity at all.
• Curtailed wind and solar in Texas and the Pacific Northwest. Wind farms in West Texas and solar farms in the Southwest regularly generate power that the grid cannot absorb during low-demand periods. Rather than curtailing (shutting off) the generation, some operators have co-located mining operations that consume the surplus and can shut down instantly when grid demand rises. The Bitcoin Miner Council estimated that approximately 50–60% of the mining capacity in the ERCOT (Texas) grid is associated with some level of curtailed or stranded renewable energy.
• Behind-the-meter industrial surplus. Some industrial facilities — paper mills, aluminum smelters, large manufacturing plants — have moments of surplus generation from on-site power systems. On-site mining can absorb this surplus without requiring a grid connection.

The honest qualification is that “stranded energy” mining is a small fraction of total Bitcoin mining. Most mining occurs at grid-connected facilities in competitive electricity markets, drawing from the same grid as homes and businesses. The stranded energy argument is real for a subset of the industry; it is not a description of how most of the hashrate is powered.

Flare Gas Monetization

This is the most environmentally interesting application of cryptocurrency mining, and it deserves careful treatment because the benefit is real but the framing is frequently exaggerated.

Oil and gas production generates associated natural gas that, at remote wellheads without pipeline infrastructure, has historically been either vented directly to the atmosphere or flared — burned in an open flame. The World Bank estimates that approximately 140 billion cubic meters of natural gas are flared globally each year — wasted energy with significant environmental consequence. Methane (the primary component of natural gas) has a global warming potential approximately 80 times that of CO₂ over a 20-year timeframe. Flaring converts most of the methane to CO₂, which is less harmful but still emits carbon.

Companies like Crusoe Energy have built a business around deploying containerized Bitcoin mining units directly at oil and gas wellheads, powered by gas that would otherwise be flared. The environmental math:

• Gas that is flared is converted to CO₂ with approximately 98% combustion efficiency — better than venting raw methane, but still an emission
• Gas that powers a mining generator is also converted to CO₂, but approximately 30–35% of the energy is captured as electricity to run the miners
• The effective emission per unit of gas is similar to flaring, but the mining operation extracts productive use from the energy before it is released
• More importantly, some flare gas mining replaces gas that would have been vented rather than flared — where raw methane would otherwise have been released, converting it to CO₂ via combustion is a genuine greenhouse gas reduction

Crusoe Energy published data suggesting its operations reduce CO₂-equivalent emissions by approximately 63% compared to flaring. Independent analysis has found smaller but positive benefits in scenarios where flaring actually occurs. The important caveat: flare gas mining is only environmentally beneficial if the alternative is actually flaring. There is evidence that the economics of mining have in some cases induced gas production that would not otherwise have occurred, or diverted gas from productive pipeline use to lower-value mining. The net environmental benefit is real in genuine stranded-gas applications and questionable in others.

Demand Response and Grid Flexibility

The most underappreciated legitimate value that large-scale mining provides to power grids is controllable demand. Unlike most industrial loads that require advance notice to curtail, mining operations can reduce consumption to near zero within seconds — simply by stopping the miners. No process is interrupted, no product is damaged, no customer is harmed. The miners pause, the grid gets its power back, and mining resumes when conditions allow.

This is exactly the type of flexible demand that grid operators need as variable renewable generation increases grid volatility. ERCOT (the Texas grid) has formalized this relationship: multiple large Bitcoin mining operations are registered as Controllable Load Resources, meaning they participate in ancillary services markets and receive payments for being available to curtail on demand.

During the February 2021 Winter Storm Uri, Bitcoin miners in Texas voluntarily curtailed approximately 1,500 MW of load to free up grid capacity during the crisis. During extreme summer heat events in 2022 and 2023, Texas miners again curtailed on ERCOT request, providing meaningful relief during periods when grid reserves were thin. This is a real and valuable grid service — equivalent to operating a dispatchable power plant in reverse — and it is something that most industrial loads cannot provide at comparable speed or scale.

The honest limits of this argument: demand response value is proportional to how often curtailment is called and how much miners are compensated for it. In most markets, this represents a modest fraction of mining revenue. It is a genuine benefit but not a sufficient environmental offset for the baseline energy consumption during the other 8,000+ hours per year when the miners run at full load.

The Bad: Real Costs That Scale With the Network

The Carbon Footprint in Honest Numbers

The 22–30 million metric ton annual CO₂ estimate for Bitcoin mining is a real number and it scales with price. When Bitcoin’s price rises, mining becomes more profitable, more miners deploy hardware, network difficulty rises, and energy consumption increases until profitability returns to equilibrium. There is no mechanism within the proof-of-work design that caps energy consumption regardless of how efficient the hardware becomes or how many miners are online.

A direct comparison that illustrates the problem:

• Bitcoin network: processes approximately 7 transactions per second, consumes ~140 TWh/year
• Visa network: processes approximately 24,000 transactions per second at peak, consumes ~0.3 TWh/year
• Energy per Bitcoin transaction: ~1,000–2,000 kWh — roughly the monthly electricity use of an average U.S. household
• Energy per Visa transaction: ~0.001 kWh — less than running a lightbulb for a few seconds

Bitcoin advocates correctly point out that this comparison is not apples-to-apples — Bitcoin is a settlement network, not a payment processor, and second-layer solutions like the Lightning Network handle small transactions with minimal energy overhead. This is fair. It does not change the fact that the base layer of the Bitcoin network consumes extraordinary amounts of energy to process a small number of transactions, and that the energy cost scales with the network’s security budget rather than with its transaction volume.

The ASIC E-Waste Crisis

Mining hardware has an extraordinarily short productive lifespan. A top-of-the-line ASIC miner released today will likely be unprofitable within 18–36 months, not because it has broken down but because newer-generation hardware with higher hash rates and better energy efficiency per hash has been released, raising network difficulty to the point where the older machine cannot mine profitably at any reasonable electricity price.

The result is a continuous stream of discarded mining hardware that is, for practical purposes, entirely non-recyclable:

• ASICs are highly specialized chips with no alternative application. Unlike a server CPU that might be repurposed in a less demanding computing environment, an ASIC miner that is uneconomical for Bitcoin mining is uneconomical for any purpose.
• The circuit boards, chips, and packaging contain the same hazardous materials as conventional electronics — lead, brominated flame retardants, beryllium — but the specialized nature of the hardware means it cannot enter standard refurbishment pipelines.
• A 2021 study published in Resources, Conservation and Recycling estimated that the Bitcoin network generates approximately 30,000–35,000 metric tons of e-waste annually. That is comparable to the e-waste generated by the Netherlands — a country of 17 million people — from all its consumer electronics combined.

The halving mechanism makes this worse in a specific way. Every four years, Bitcoin’s block reward is cut in half. After each halving, miners operating on marginal electricity prices are suddenly operating at a loss; the only response is to find cheaper power or retire the equipment. The April 2024 halving (reducing the reward from 6.25 to 3.125 BTC) forced a significant wave of equipment retirements as older-generation ASICs became unviable at any commercially available electricity rate. This pattern will repeat with every subsequent halving.

Community Electricity Impacts

When a large mining operation enters a small electricity market, the impact on local rates and grid reliability can be significant and largely uncompensated. Several documented cases illustrate the pattern:

• Plattsburgh, New York (2018): The city’s low-cost hydroelectric power allocation was exhausted by Bitcoin miners, forcing the city to purchase expensive power on the spot market. Residential electricity bills rose by $100–$200/month for thousands of customers. The city enacted an 18-month moratorium on new crypto mining permits — one of the first such actions in the U.S.
• Kentucky and Georgia: Rural electric cooperatives in both states have reported that large mining facilities have consumed power allocations intended for economic development in their service territories, crowding out manufacturers and processors that would have employed local residents.
• Kazakhstan: The rapid influx of mining operations after China’s 2021 ban contributed to grid instability and blackouts across the country during the winter of 2021–2022, leading to government-ordered curtailments and subsequently a crackdown on unregistered mining.

The underlying dynamic is that mining is highly mobile and extremely price-sensitive. Miners go where electricity is cheapest. The communities with the cheapest electricity are often those where local economic development has been limited and power infrastructure was built for industrial or utility loads that never fully materialized. When mining fills that capacity, it crowds out local uses and, if it leaves as quickly as it arrived, leaves communities with rate increases and no lasting economic benefit.

The Ugly: The Structural Problems That Cannot Be Fixed

The Purposeful Futility of Proof of Work

The most honest description of what Bitcoin mining does is this: it performs a cryptographic lottery. Billions of times per second, miners generate random numbers and check whether a mathematical condition is satisfied. When one miner happens to find a number that satisfies the condition, they win the block reward. The computation performed in the lottery has no value except the lottery itself — the hash outputs are discarded, the work cannot be reused, and the process starts over for the next block.

This is not a design flaw that engineers failed to notice. It is an intentional design choice. Satoshi Nakamoto’s original Bitcoin whitepaper explicitly describes the energy expenditure as the security mechanism: making fraud expensive enough to deter — since rewriting history on the blockchain requires re-doing all the proof-of-work from the target block forward, which would cost more in electricity than any realistic fraudulent gain. The energy waste is the product. Without it, the security model collapses.

This distinguishes crypto mining from other industrial energy consumers in an important way. Steel production uses enormous energy but produces steel. Aluminum smelting uses enormous energy but produces aluminum. Data centers use enormous energy but provide computing services that run the digital economy. Bitcoin mining uses enormous energy and produces Bitcoin — a digital token whose value exists entirely as a social consensus about its value. Whether that is a worthwhile exchange is a values question, not a technical one. What is not debatable is that the energy expenditure produces nothing independently useful, and this is fundamental to the design rather than an inefficiency to be optimized away.

The Efficiency Treadmill

Semiconductor efficiency has improved dramatically over the past decade. The energy consumed per terahash (TH) of Bitcoin mining computation has fallen by roughly 99% since 2013, as hardware has improved from 100,000 J/TH to under 20 J/TH for current-generation ASICs. On its face, this looks like extraordinary environmental progress.

The network’s total energy consumption has not declined. It has increased every year that Bitcoin’s price has risen, because the difficulty adjustment mechanism ensures that energy consumption tracks to the block reward value rather than to hardware efficiency. More efficient hardware lowers the cost per hash, which makes mining more profitable, which attracts more miners and more hardware, which raises difficulty, until the cost of mining once again approaches the value of the reward. The efficiency improvement is fully offset by the resulting increase in deployed hashrate.

This is a fundamental property of proof-of-work that cannot be engineered around. The only mechanisms that reduce the network’s total energy consumption are a sustained decline in Bitcoin’s price (which makes mining unprofitable), a change to the consensus mechanism (which Bitcoin’s community has consistently rejected), or an externally imposed energy cost (regulation, carbon pricing). Efficiency improvements alone cannot do it.

The Geography of Dirty Power

Mining operations optimize for the cheapest electricity, not the cleanest. These are not the same thing. The cheapest electricity in many markets is produced by old, fully depreciated fossil fuel plants with low marginal operating costs. The geographic distribution of mining follows cheap power with little regard for carbon intensity, and the results are visible in the data.

Kazakhstan became the world’s second-largest mining hub after China’s 2021 ban specifically because of its cheap coal-fired electricity. The grid carbon intensity of Kazakhstan’s electricity is approximately 700–800 g CO₂/kWh — among the highest in the world for a large economy. Mining operations that relocated from China’s increasingly hydro-dominated southwest to Kazakhstan moved from a relatively clean power source to one of the dirtiest. This is the direct opposite of the “mining encourages renewable energy” narrative, and it happened at scale.

In the United States, Kentucky has attracted substantial mining investment based on its coal-powered electricity. The state’s grid carbon intensity is approximately 650 g CO₂/kWh. A 100 MW mining facility in Kentucky generates roughly 570,000 metric tons of CO₂ per year — the equivalent of adding 120,000 gasoline-powered cars to the road.

The renewable energy claims circulated by mining industry associations deserve the same scrutiny applied to data center claims in the previous article. The Bitcoin Mining Council’s estimate that 54% of Bitcoin mining uses sustainable energy is based on voluntary self-reporting by a subset of miners that represent less than half the network’s hashrate. Independent analysis by the Cambridge Centre for Alternative Finance and other researchers produces significantly lower estimates of renewable penetration in the network’s energy mix.

Electricity Theft and Developing-World Impacts

In countries with subsidized or unreliable electricity systems, crypto mining creates an additional category of harm that receives little attention in Western industry coverage. Documented patterns:

• Iran: Cheap subsidized electricity attracted enormous mining investment. The government has intermittently banned mining to protect the domestic grid from overload, with industrial and residential blackouts attributed in part to mining demand. Despite repeated bans, mining has continued through informal channels.
• Venezuela and Argentina: Hyperinflation makes cryptocurrency attractive as a store of value, which also incentivizes local mining. Subsidized electricity rates intended to support households and small businesses are consumed by mining operations at rates far above what the subsidy system anticipated. The subsidy effectively transfers public resources to private mining profits.
• Electricity theft: In parts of Central Asia, the Middle East, and Latin America, mining operations have been documented connecting directly to electrical infrastructure without metering or payment. The scale is difficult to quantify but represents a real and recurring enforcement problem for grid operators.

The Ethereum Lesson: What Proof of Stake Proved

On September 15, 2022, the Ethereum network completed “The Merge” — a transition from proof-of-work to proof-of-stake consensus. The event is the most important data point in the environmental debate about crypto mining because it demonstrated, at scale and in production, exactly what is possible.

Before The Merge, Ethereum was the world’s second-largest proof-of-work mining network, consuming an estimated 75–100 TWh per year — comparable to Chile or Austria. After The Merge, Ethereum’s electricity consumption dropped to approximately 0.01 TWh per year. The reduction was 99.95%. Overnight.

The network continued to function. Transactions continued to be processed. The blockchain continued to be secured. The price of Ether was not materially affected by the transition. What disappeared was the energy expenditure.

Proof of stake replaces computational work with economic stake as the security mechanism. Validators lock up ETH as collateral; if they try to validate fraudulent transactions, their stake is “slashed” (destroyed). The security model relies on the cost of acquiring and risking a large stake rather than the cost of burning electricity. For a network processing the same transaction volume as Ethereum, the energy requirement drops from a national-electricity-scale industrial operation to roughly the equivalent of running a few thousand laptops.

Bitcoin advocates have several responses to the Ethereum precedent, and some of them are substantive:

• Proof of stake changes the security model fundamentally. In PoW, attacking the network requires acquiring and operating physical hardware — a real-world constraint that is difficult to hide and slow to scale. In PoS, an attacker only needs to accumulate enough stake, which is a purely financial operation. Whether PoS is “as secure” as PoW is a genuine technical debate, not settled by the Ethereum transition alone.
• Bitcoin’s design philosophy treats energy expenditure as a feature. The argument is that the physical, real-world cost of securing the network — electricity, hardware, heat — creates a connection between the digital and physical worlds that pure financial staking does not. This is an ideological position, not an engineering constraint.
• Ethereum has not proven long-term PoS security at Bitcoin’s value scale. Ethereum’s market cap is substantially smaller than Bitcoin’s. Whether PoS security holds against a well-resourced attacker targeting a network worth trillions of dollars has not been tested over a long time horizon.

These arguments may or may not be persuasive depending on your view of the Bitcoin network’s purpose and value. What the Ethereum transition proved beyond dispute is that the energy consumption of a proof-of-work network is not a technical inevitability — it is a design choice. Bitcoin’s energy consumption exists because the Bitcoin community has chosen it to exist. That is a legitimate choice. It should be presented as a choice, not as a natural law.

What Can Actually Be Done

Demand Response Formalization

The Texas ERCOT model — registering large mining operations as Controllable Load Resources and integrating them into ancillary services markets — should be replicated in every grid where significant mining capacity exists. This does not reduce mining’s baseline energy consumption, but it converts what would otherwise be an inflexible load into a grid asset that provides real value during stress events. Mining operations should be required to participate in demand response programs as a condition of large-load interconnection agreements, rather than on a voluntary basis.

Flare Gas Standards and Verification

Flare gas mining is environmentally beneficial only when the alternative is genuinely flaring or venting. An environmental standard for flare gas mining should require third-party verification that the gas being consumed would otherwise be wasted — not diverted from pipeline production, not used to incentivize additional drilling, and not displacing other beneficial uses. Companies like Crusoe Energy that operate legitimately in genuine stranded-gas applications should be able to demonstrate this verification. Operations that cannot should not receive the environmental credit.

Carbon Pricing and Energy Disclosure

The most direct mechanism for internalizing crypto mining’s environmental cost is carbon pricing. If electricity prices reflect the carbon cost of generation, mining operations in coal-heavy grids become less competitive relative to those in clean-energy markets. Miners would have a direct financial incentive to locate where renewable energy is genuinely available rather than claiming credit for RECs purchased elsewhere. Most major economies have not implemented carbon pricing that would affect electricity-intensive industries at the scale necessary to materially alter mining location decisions.

Mandatory energy source disclosure for registered mining operations — the equivalent of the renewable energy disclosure requirements being developed for data centers — would at minimum create public accountability for the network’s actual energy mix, which is currently estimated from indirect inference rather than direct measurement.

Waste Heat Recovery

Mining facilities generate consistent, high-volume waste heat that is currently rejected to the atmosphere in virtually all cases. Opportunities for productive use exist at smaller scales than typical data center waste heat programs:

• Greenhouse heating: several small mining operations in cold climates have redirected waste heat to agricultural greenhouses, extending growing seasons at near-zero heating cost
• Aquaculture: fish farming requires stable water temperatures; mining waste heat has been used to maintain tank temperatures in cold climates
• Industrial drying: low-grade heat (40–60°C) from mining cooling systems can be used for industrial drying processes in food and materials manufacturing

These applications are niche and do not offset a material fraction of mining’s total energy consumption, but they represent real value recovery where proximity and application align.

The Option Bitcoin Will Not Take

Ethereum proved that a major cryptocurrency network can transition from proof-of-work to proof-of-stake without destroying its functionality or security. The option is available to Bitcoin. The Bitcoin development community and the mining industry have consistently rejected it and will continue to do so. The mining industry’s rejection is straightforwardly self-interested: miners have invested billions in ASIC hardware whose only purpose is Bitcoin proof-of-work. A transition to proof-of-stake would make that investment worthless overnight. The philosophical rejection among Bitcoin maximalists is more principled, if debatable.

The honest conclusion is that Bitcoin’s proof-of-work energy consumption will not be significantly reduced by anything short of a sustained price collapse, external regulation, or a fundamental change in the network’s consensus mechanism. None of those outcomes appears likely in the near term. The energy consumption should be understood as a permanent feature of Bitcoin at current and higher price levels, not a temporary inefficiency being addressed.

Bottom Line

Is crypto mining as bad as critics claim?

On energy consumption: largely yes. The scale is real, the carbon intensity is variable but substantial in aggregate, and the efficiency treadmill ensures that hardware improvements do not reduce total consumption — they raise difficulty until the consumption returns to economic equilibrium. The standard critique is not exaggerated.

The honest qualifications that cut the other way:

• Stranded energy applications — genuine surplus hydro, remote flare gas, curtailed renewables — represent a real and legitimate use case where mining provides environmental value that would not otherwise exist. This is a minority of total mining activity but it is not nothing.
• Grid demand response is a genuine and undervalued service. The Texas model demonstrates that mining can provide meaningful grid stability support in ways that most other industrial loads cannot.
• The comparison to gold mining is not entirely unfair. The World Gold Council estimates gold mining consumes approximately 100–130 TWh/year globally, produces significant land disruption, uses toxic chemicals, and generates tailings waste — for a metal whose primary modern application is jewelry and financial reserves. Whether gold or Bitcoin is a more justified use of that energy is a values question with no obvious correct answer.

The honest accounting

Dimension	Reality
Annual energy consumption (Bitcoin)	~120–150 TWh — comparable to Argentina
Annual CO2 emissions	~22–30 million metric tons
Energy per base-layer transaction	~1,000–2,000 kWh — a month of household electricity
Annual e-waste generated	~30,000–35,000 metric tons; essentially zero recyclable
Renewable energy share (est.)	~25–40% (disputed; industry self-reports higher)
Efficiency improvement potential	Zero net effect — difficulty adjusts to consume gains
Stranded/flare gas benefit	Real but represents minority of total hashrate
Grid demand response value	Genuine; 1,500 MW curtailed in Texas during Uri
Net environmental verdict	Significant liability; marginal benefits do not offset

Ethereum’s transition to proof-of-stake is the most important fact in this debate. It demonstrated that the energy consumption of a major cryptocurrency network is a design choice, not a physical necessity. Bitcoin’s community has made a different choice, and they are entitled to make it. What is not accurate is presenting that choice as inevitable, as environmentally neutral, or as primarily driven by a genuine commitment to clean energy.

Bitcoin mining consumes as much electricity as a mid-sized country to run a lottery. The lottery has value to some people. That does not make the electricity free.