Original paper:
Introduction:
Gill, Stinner, and Tyrell, through their paper, make the claim that utilising otherwise wasted renewable energy by mining Bitcoin inhibits, rather than increases, the kinds of upgrades and evolutions that a decarbonised energy system needs. They claim that increases in mining actually increase the impact of certain negative externalities, in particular CO₂ and e-waste. However, we argue, in refutation of these claims, that actually the opposite is not only true, but far better supported by evidence from practice. Bitcoin mining can complement existing and decarbonising energy systems, support the stability of grids, and aid in the deployment of new renewable capacity. Gill et al. use a cast of usual suspects of debunked and out-of-date sources for their data and information—in particular, the oft-cited de Vries, and Mora, and the Cambridge Bitcoin Electricity Consumption Index. They also fail to include crucial protocol-level factors such as difficulty adjustments and halving cycles, and positive externalities such as financial inclusion, hedging inflation, and resilience in volatile economies. Consequently, the conclusions in their article are found from incomplete modelling which ignores both real-world evidence and good practice in economic research, which paints a wholly different picture of what Bitcoin’s role in modern energy systems both is, and could be.
Generally Speaking: Poor Basis, Outside Context Claims
One of the paper's key arguments is that using otherwise wasted renewable energy for mining risks crowding out more socially beneficial investments in grid capacity or energy storage. The authors state: "Monetizing surplus energy through Bitcoin mining creates counter-incentives for investing in storage and grid infrastructure." This claim is examined by these researchers through an equilibrium model; however, the literature review uses outdated data in the form of the Cambridge Bitcoin Electricity Consumption Index’s data from 2022, and methodologically impoverished literature such as Alex de Vries’ entirely fabricated and debunked energy and e-waste guesstimates. The fact of the matter is that the empirical evidence demonstrates findings that are entirely the opposite. A recent whitepaper by the team at the Nicholas Institute for Energy, Environment & Sustainability from Duke University recently demonstrated that Bitcoin mining has been clearly shown to help grid operators avoid the need for grid upgrades. Other studies show that grid stability services such as demand response are Bitcoin mining’s killer app. While ARK Invest (2021) showed that battery storage and Bitcoin mining are not competitive, instead, storage and Bitcoin mining work synergistically together as a means of arbitraging energy revenues while also providing grid stability. Other studies have shown that Bitcoin mining both incentivises the deployment of new renewable generation capacity, and also enables what would otherwise be marginal renewable energy generation to be financed and built.
The inexperience with Bitcoin mining of Gill et al. is clear in much of the article. Often concepts are sort of thrown together in both a confused and misleading way. For example, readers may find comments like the following fairly hard to follow due to the fragile nature of the links between the concepts: “…miners’ dependency on Bitcoin’s volatile price together with their mobility makes them unreliable partners for the energy transition”. Firstly, the statement is wrong on both unrelated points. Location flexibility is exactly the reason miners are able to co-locate with renewable generation facilities, something widely established in both case studies and peer reviewed research. This rebuttal will discuss this point in more detail later, but it’s worth noting here that while miners are flexible about where they can operate, they cannot teleport exahashes of capacity to the newest, cheapest electricity source. As for price volatility, this is a moot (and frankly highly misleading) point, because Bitcoin miners don't pay generators in Bitcoin, they pay them in fiat currency. Volatility in the price of Bitcoin is actually a major pain point for miners because it makes the calculation of potential returns on any investment in both OPEX and CAPEX hard to predict. Speaking from experience, it’s far more likely miners have actually missed out on growth funding or passed on energy opportunities because of price volatility than have somehow leveraged it for some unscrupulous benefit.
The article argues that there is a pressing need for an economics-based appraisal of Bitcoin and its externalities due to a lack of examination through these kinds of methods in the past. The authors state:
"The arguments of each side, however, originate from fields outside of economics, commonly missing essential dynamic effects for a conclusive assessment. This paper formulates the question in economic terms, providing policymakers with insights into the key trade-offs and long-term impacts."
This is, however, incorrect, and demonstrates a significant failing on the part of the authors. That is, they have failed to conduct a sufficient literature review. There are many research papers on Bitcoin mining, including from economists, and those which use an economics-based approach. Murray Rudd, co-author of multiple Bitcoin papers, applies an economics lens to his writing and is from an economics background. However, he also is deeply knowledgeable in both Environmental Science, Environmental Economics, and Bitcoin mining. One of his papers performed a Monte Carlo analysis to show that Bitcoin mining was uniquely suited to profitably mitigate methane emissions from landfills. Jan Wüstenfeld, researcher from the Digital Assets Research Institute has written extensively on Bitcoin mining and applies an economics lens to his research. Michael Kazakka authored a paper from an economics lens on Bitcoin, examining its environmental impact with that of the traditional finance sector.
Clearly, the claim that arguments (both for and against) are insufficiently examined from an economic perspective is entirely unfounded. The analysis by Gill et al. may be unique, but it is not a uniqueness that is advisable in writing about a domain that is at its core multi-disciplinary. Indeed, far from being an advertisement for the benefits of an economist's lens on Bitcoin, the paper is an inadvertent warning to others of the dangers of having a purely economic lens on writing about Bitcoin mining, to the exclusion of other germane domain knowledge. The fact is, by excluding so much of what can be captured in a multidisciplinary analysis, the results and thus the claims made by the authors are wildly inaccurate.
However, a lack of domain knowledge doesn’t stop the authors from making some sweeping statements about said domain. The claim: "It is no understatement to say that this network design burdens substantial externalities on society.” The externalities the paper claims are first: CO₂ emissions, and second, e-waste. Both of these externality driving factors are widely discredited in literature. Their key references – de Vries and Mora have built a significant basis of citations on their damning commentary on Bitcoin. They claim, both independently and together, that Bitcoin mining generates vast amounts of CO₂ emissions, and tonnes of e-waste in discarded mining computers. The problem is that none of it’s true. Bitcoin, like EVs have no direct emissions, and is the most demonstrably sustainably powered industry in existence at over 56% renewable energy consumption. While claims about e-waste are even more spurious. Not only is the second-hand market for mining computers incredibly robust, the miners themselves are 99% entirely recyclable - and profitably too.
Negative Externalities Only
Also worthy of noting is that the paper focuses exclusively on the negative externalities of Bitcoin mining, in particular the CO₂ and e-waste that mining can produce (indirectly in the case of CO₂). Despite the flawed figures that the paper uses to base its analysis, it also fails to include any discussion of positive externalities that might stem from mining, or even the existence of the Bitcoin network. If the paper is to ultimately argue that Bitcoin is harmful to society, then a cost-benefit lens should be applied to at least mention the potential upsides. These could include, but are by no means limited to, financial inclusion, censorship resistance, innovation in financial products and services, the enrichment of owners, cost savings in other areas, and the reduction of the use of resource-intensive legacy systems. Bitcoin mining is being used to heat tens of thousands of homes in Finland, and in so doing saving hundreds of thousands of tonnes of greenhouse gasses from being emitted. Bitcoin mining monetises microgrids in Africa, enabling economic development and electrification of small rural villages. By failing to contrast these benefits to the negative externalities and continuing to conclude a net-negative impact, the paper could never have arrived at any other conclusion. The outcome was tautological.
To be clear, because the methods are so flawed, these authors have failed to prove their claims, have used incorrect input values for calculations, and demonstrated a lack of understanding of the field. Having only examined Bitcoin from the perspective of two externalities: those of CO₂ and e-waste they have missed so much of the full picture one can only look askance at these results. There are at least 19 positive externalities that are both well documented and empirically demonstrated. Many of which aside from the already listed above such as the prevention of wealth erosion in countries with hyperinflation, providing banking-like services to the underbanked, and enabling refugees to retain their wealth and rebuild their financial independence after displacement, have an economically measurable basis. Despite this, these externalities are not included in the analysis. One wonders just how someone prepared to do so much work on Bitcoin could miss so much of what makes up the economics of it.
Garbage In, Garbage Out: When Even Good Maths Can’t Save You
The paper’s poor basis for the inputs in the study, flawed assumptions, and generally biased approach to analysis necessarily results in an article that results in flawed findings. The following is a discussion of how the inadequate literature review and holistic investigation of externalities that went into the analysis has resulted in a dearth of insight that risks disseminating further misinformation into the public discourse.
1) Surplus Renewable Energy Reduces Emissions but Increases E-Waste
Paper Result: The first finding for the paper is that cheap “otherwise wasted” energy from renewables displaces fossil sources, lowering CO₂ output. However, this same cost advantage encourages more mining rigs to enter, leading to higher e-waste.
Ignores multi-period feedback
Putting aside the fact that e-waste is demonstrably overstated as an externality due to the vanishingly small proportion of the hardware that’s actually truly wasted when reaching end of life after a long career on the racks of datacenters, the paper makes a number of errors and mischaracterisations that diminish the impact and accuracy of their findings. The paper focuses on a single-period, static equilibrium. This neglects just how dynamic the bitcoin mining protocol and therefore market is in reality. For example, if new miners appear on the network due to some significant new sources of cheap energy, mining difficulty adjustments reduce per-rig mining returns. This feedback dampens or negates Gill et al.'s hypothesised leap in the number of active devices and thus limits their surge in anticipated e-waste. It would also have the counter-effect of (even) further reducing emissions—because miners whose energy has a higher marginal cost per kWh would be forced to exit the business. Fossil fuel generators have the highest marginal costs and as such are less attractive to miners for reasons of cost and volatility. So, shifting to cheap energy is far more likely to both reduce emissions and moderate e-waste, in contrast to the findings of Gill et al.
Understates regional constraints
In reality, surplus renewable energy is often location-specific and time-bound. Local grid limitations or local regulations can constrain how many miners can actually be deployed in any location. The authors’ framework treats this surplus as perfectly accessible once available, which under-exaggerates the barriers to entry for new mining rigs (and resulting e-waste). The pre-conditions to build a data centre can include the legality, duration of energy availability, political stability, serviceability, staffing, size, location, permitting and consenting, accessibility, ground readiness, among others. The fact is that the lead time for building a new Bitcoin data center extends from a bare minimum of three months to multiple years. Bitcoin mining is not like water that flows to the lowest point. It’s a marble run of conditions that need to be satisfied before it can begin operating.
Alternate fates for hardware
The paper assumes all rigs go from “active” to “discarded.” Yet some rigs are sold on secondary markets, repurposed in regions with even cheaper power, or run intermittently for load balancing. These real-world practices may change actual e-waste profiles in ways the paper’s simple analysis and characterisation misses. We know from both peer-reviewed research and secondary markets that even very old hardware has a significant lifespan. Gill et al. rely on de Vries’ roundly debunked estimates of 1.56 years for the lifespan of a mining computer. The reality is they last at least four times longer than that, and in many cases far, far longer. The S9 series of miners, for instance, has found a secondary (or tertiary depending on the unit and the number of hands it has passed through) home in being repurposed as self-subsidising heaters for homes and offices.
Real-world fraction of fossil fuel displacement
The finding that surplus renewables “displace fossil energy” presumes miners otherwise would have used more carbon-intensive power. This counterfactual is not always well-defined. In regions like Texas, growth in both renewable and fossil power have occurred simultaneously. If surplus consumption by miners is partially substituting gas peaker plants or is just layering onto total demand, the net effect is more ambiguous. It’s well documented that Bitcoin mining gravitates to the energy with the lowest marginal cost. Because renewables have an effective marginal cost of $0/kWh, miners would always opt for energy at this source rather than from fossil fuels. Gill et al. ignore marginal energy costs, which is an issue because fossil fuels are priced on commodity markets making the actual marginal cost variable and hence unattractive to miners. Another failure in the analysis is the apparent ignorance to the fact that peaker plants are employed when demand from local retail consumption is at its highest, and thus, when miners are the least incentivised to operate. The idea that mining drives the use of peaker plants or increased fossil fuels due to higher overall grid demand where miners are present is again, based on a flawed assumption.
As such, while it’s plausible that cheaper energy can drive more mining rigs into operation, the paper’s quantitative claims about the resulting magnitude of e-waste and exact offsetting of emissions would clearly be overstated due to their simplified, static approach and assumptions about perfect substitution. As discussed above their assumptions about the actual volume of e-waste produced is also wildly overstated, meaning any maths that use these variables as inputs produces a carried error that renders any findings on this flawed from the start.
2) Carbon Taxes Are Effective Only if Adopted Globally
Paper Result: The paper claims a Pigouvian tax (that is a tax on a transaction that creates a negative externality) on fossil-based electricity, if uniformly applied, lowers total externalities (both CO₂ and e-waste). If it’s unilateral, it spurs “carbon leakage” by pushing miners to untaxed regions, potentially raising global emissions and hardware use.
Elasticity oversimplification
Again, for this finding the authors need to assume perfect mobility of miners across borders and high elasticity of fossil supply. In reality, location-specific constraints like regulatory hurdles, political risk, or insufficient power infrastructure may limit large-scale relocation. As above, the idea of freedom of movement of miners is a wholly flawed assumption. Additionally, there’s no reason to assume that fossil fuel producers would instantly drop prices for non-Pigouvian regions. Firstly, new demand would suggest they could maintain higher prices. Secondly, they often can’t meaningfully change their prices due to marginal costs that are higher than renewables. Furthermore, grid constraints and the cost that must be borne for miners to access the energy network for example will quickly make rushing to install mining in countries like Australia with its abundant solar electricity a non-starter. Permitting and consents for construction will delay the deployment of new mining capacity for months and even years. Political risk often comes hand in hand with low electricity prices.
Partial success of unilateral measures
The authors’ conclusion that unilateral taxes are worse than no tax presupposes that the entire mining pool flees to cheaper fossil-heavy grids. But some jurisdictions might have additional incentives (e.g., renewable subsidies, stable policy environment) that keep or attract miners even with a carbon tax. This partial shift could still yield net emission declines within taxed regions and, there’s no reason to assume that this would, or would not, proportionally increase mining activities elsewhere.
Political economy complexities
The authors treat global adoption as theoretically straightforward, but in practice, coordinated global carbon policy is extremely challenging. It’s possible that even partial adoption can spur a long-term shift in corporate norms, hardware design, or consumer preference, lessening the severity of the predicted “leakage.” As such, while the paper’s logic on carbon taxes follows, the claim that unilateral taxes definitively fail to cut global externalities relies on assumptions about perfect miner mobility and uniform power pricing. Real-world frictions effectively invalidate the “leakage is inevitable” conclusion.
3) Long-Term Energy Transition Is Hindered by Surplus Monetisation
Paper Result: The authors claim that by monetising surplus renewable energy through mining, generators and grid operators reduce their incentive to invest in grid and storage expansions. This delays or hinders a full-scale transition away from fossil fuels.
Limited empirical basis
The authors present only theoretical reasoning for “crowding out” grid or storage infrastructure. Actual investment decisions are shaped by policy mandates, capacity payments, regulated returns on infrastructure, and local market structures. In many regions, utility companies are required to expand grid/storage capacity irrespective of whether they can earn side revenues from miners (see: EU Electricity Directive 944, Regulation EC No. 714, Federal Power Act, FERC Order No. 1000, among others). Low priced energy during peak renewables generation hours and/or low retail demand periods is better off stored in batteries for release during higher yielding timeframes. Curtailed energy is zero-priced but during peak demand that same electricity may be worth hundreds of dollars per megawatt. Even Bitcoin miners with their ability to absorb instantly and at high capacities cannot compete with retail demand prices. So where these authors offer limited proof for their claims, the real world offers longitudinal real world evidence to the contrary.
Counter-argument: bridging cost gaps
If an energy project’s revenues are too low without monetising surplus, it might never be built in the first place. There is already very good evidence that selling surplus electricity to miners can bridge an investment hurdle or accelerate the time-to-payoff for certain renewable projects. Over time, more robust renewable capacity could lead to an overall higher fraction of carbon-free supply. Furthermore, there is a counter-scenario that the paper fails to examine: once these projects are profitable and financed, operators might then build the storage or grid expansions anyway, having gained capital from the extra sales of energy to miners.
Different investor profiles
Grid or storage expansions often involve large-scale utilities, government bodies, or specialised investors while hosting miners is typically done by private power-plant operators looking for quick returns. These are not always the same decision-makers or the same capital sources. So it’s not always a direct either-or-trade-off. In fact it’s well established that battery storage and mining are complementary rather than competitive. It’s important to note the fact that batteries can only be charged once - and then must be discharged before they can be charged again. As such the installed capacity for battery storage is generally relatively small when compared with the total capacity at the generator generally in the range of 10 - 25% of the power plant capacity or about 1–4 hours relative to the plant’s full output. This is because battery storage is designed to smooth out short term demand fluctuations, not act as a power plant in its own right. Meanwhile miners can monetise any remaining wasted energy, and can be used to discharge batteries for either battery health management, or in preparation to recharge using surplus energy for the next cycle.
Neglect of concurrent regulatory incentives
Many countries have renewable energy targets, feed-in tariffs, or storage incentives that drive new infrastructure. These policies are generally nationwide, and as such, their total impact on both generators and utility operators could easily dwarf the incremental revenue from Bitcoin mining, making it less likely that monetising surplus energy truly deters larger, systemic upgrades.
Methods, Not Madness: Or Why Being Boxed In Leaves You In The Dark
The article is based in the paradigm of positivist economics. That is, the authors acknowledge that there exists an objective reality in which the phenomenon under study can be measured and reflected in the results of that study. In economics, the method of choice for these representations is a generally accepted group of models that can, and have in the past, been used to represent reality by defining the inputs, assigning them to terms in the model, and running the calculations to produce an outcome. The idea is that this outcome tells us something about the nature of the phenomenon under study. The issue is that unless the researcher is working with good information, good methods, and good intentions, you don’t tend to get good outcomes. In this case, by using already debunked inputs to the study—in particular information that either comes from or is itself informed by authors like de Vries and Mora—the methods can only deliver the derivations of the poor inputs.
The paper’s model is essentially a partial-equilibrium representation of Bitcoin mining, treating block rewards (revenues) as exogenous—that is, they are introduced from outside the system—and then deriving costs and resource usage in a single-period setting. Partial-equilibrium approaches can be appropriate for highlighting certain trade-offs (e.g., surplus-energy usage versus hardware cost), but they also risk oversimplifying how the real-world mining market evolves over time.
The authors also ignore the way the protocol works to moderate its own operation. Their model does not incorporate the halving, or potentially large swings in transaction fees and bitcoin price. Over medium or long horizons, both the block reward and miners’ revenue composition (fees vs. new Bitcoin) change dramatically, so any extrapolation of emissions or e-waste from a single static snapshot is by nature misleading. In particular, the network difficulty is adjusted to keep block times roughly constant at every 10 minutes. If new miners flood in, this difficulty rises, reducing per machine returns and dampening the business case to invest in new hardware. Gill et al.’s simplification introduces significant bias in estimates of how quickly mining activity can or will expand when new cheap energy appears. To be clear, failing to include variables like a difficulty adjustment and miner returns is not outside the capability of this kind of research. Budish, 2024 shows that it’s possible to include difficulty adjustment in an economic analysis. Pratt & Walter conducted an equilibrium model analysis for their paper, and again demonstrated that increased hardware input can raise difficulty, affecting equilibrium resource usage. So failing to build on the methodological basis of other established literature exposes them as either lazy, ill-informed, or both.
Summary
This article by Gill, Stinner, and Tyrell is yet another paper in what is becoming a long-standing tradition of criticising Bitcoin from angles and perspectives which both fail to critically examine the quality of the references they’re extrapolating from, and conduct incomplete analysis. It has become clear, at least to this author, that a lack of expertise in the domain under study leads necessarily to flawed thinking, flawed research, and flawed outcomes. As a result, they are fairly simple to rebut. The unfortunate thing is that it takes time to do so.
The rebuttal here argues that the claim made by Gill et al. regarding Bitcoin miners “crowding out” investments in grid capacity is purely hypothetical and unsupported by real-world evidence. In fact, studies from Duke University’s Nicholas Institute and ARK Invest indicate the exact opposite—that Bitcoin miners can help balance grids through demand response and can complement storage solutions, not replace or discourage them. Furthermore, the paper’s authors rely heavily on sources (e.g., de Vries and Mora) whose estimates on Bitcoin’s energy use and e-waste have been largely discredited in more contemporary research, having been found to wildly overstate the actual state of affairs. Additionally, while the original paper focuses exclusively on what it considers negative externalities (CO₂ emissions and e-waste), it ignores a host of potential benefits from mining and the Bitcoin network itself, such as financial inclusion and wealth preservation in regions with inflation or political instability.
Methodologically, Gill et al. also has significant gaps: the authors employ a static partial-equilibrium model that omits key dynamics like difficulty adjustments, transaction-fee fluctuations, and the Bitcoin halving schedule. These oversights severely limit any accuracy and relevance the study could claim, as real-world mining markets constantly evolve in response to these dynamic and impactful protocol-level mechanisms. When we consider these issues alongside the inaccuracies made in assumptions of hardware lifespan and the sweeping generalisations about miners’ perfect mobility, the result is a set of flawed conclusions that lack proper empirical grounding and disregard existing research in both economics and energy studies.
