Full length article
Bitcoin's growing e-waste problem

https://doi.org/10.1016/j.resconrec.2021.105901Get rights and content

Highlights

  • Bitcoin's annual e-waste generation adds up to 30.7 metric kilotons as of May 2021.

  • This level is comparable to the small IT equipment waste produced by a country such as the Netherlands.

  • On average Bitcoin generates 272 g of e-waste per transaction processed on the blockchain.

  • Bitcoin could produce up to 64.4 metric kilotons of e-waste at peak Bitcoin price levels seen in early 2021.

  • The soaring demand for mining hardware may disrupt global semiconductor supply chains.

Abstract

Bitcoin's increasing energy consumption has triggered a passionate debate about the sustainability of the digital currency. And yet, most studies have thus far ignored that Bitcoin miners cycle through a growing amount of short-lived hardware that could exacerbate the growth in global electronic waste. E-waste represents a growing threat to our environment, from toxic chemicals and heavy metals leaching into soils, to air and water pollutions caused by improper recycling. Here we present a methodology to estimate Bitcoin's e-waste and find that it adds up to 30.7 metric kilotons annually, per May 2021. This number is comparable to the amount of small IT and telecommunication equipment waste produced by a country like the Netherlands. At peak Bitcoin price levels seen early in 2021, the annual amount of e-waste may grow beyond 64.4 metric kilotons in the midterm, which highlights the dynamic trend if the Bitcoin price rises further. Moreover, the demand for mining hardware already today disrupts the global semiconductor supply chain. The strategies we present may help to mitigate Bitcoin's growing e-waste problem.

Introduction

Global waste is expected to grow by 70% by 2050 from 2016 (Schrader-King and Liu, 2018). In terms of the environmental impact of waste, plastics have received the most attention as microplastics in the world's oceans already outnumber the stars in the Milky Way (UN News, 2017). Though it is less discussed, electronic waste (e-waste) – which is the waste produced by discarding electrical or electronic equipment – represents a growing threat to our environment and includes issues from toxic chemicals and heavy metals leaching into soils to air and water pollution caused by improper recycling. Of the 53.6 million metric tons (Mt) of e-waste generated globally in 2019, only 17.4% was collected and recycled (Forti et al., 2020). The amount of e-waste is expected to double by 2050 (United Nations University, 2019), and this prediction does not include the effect Bitcoin mining might have. Most research on the environmental impacts of Bitcoin (and similar cryptocurrencies) has focused on energy demand and carbon emissions and has thus far ignored that Bitcoin miners cycle through a growing amount of short-lived hardware that could exacerbate the growth in global e-waste.

Bitcoin mining started with the initial release of the digital currency in 2009. Mining is an essential activity in the Bitcoin network to validate transactions and ownership that involves adding new blocks to a chain. Each mining node bundles new transactions before solving a computationally expensive puzzle to find a ‘proof-of-work’ (PoW) for a block (Nakamoto, 2008). The first miner who finds a PoW that satisfies predetermined conditions broadcasts the block to all nodes in the network. The receiving nodes express their acceptance of the new block by building on top of it. The process repeats after each block addition. The successful miner receives newly created Bitcoins and fees for transaction validation, which provide incentives to participate in the process. Given that network participants invest time and energy into extracting resources, the process resembles metal mining, hence the name adopted for the activity (Nakamoto, 2008).

The increasing energy consumption of Bitcoin mining has triggered a passionate debate within academic literature and among the general public regarding the sustainability of the digital currencies. Bitcoin mining has been found to consume as much energy as small countries, which translates into a significant carbon footprint. However, studies have charted a wide range of results, as shown in Fig. 1. Stoll et al. (2019) found annual emissions ranging from 22.0 to 22.9 million metric tons of carbon dioxide (MtCO2). Krause and Tolaymat (2018) provided an estimated range from 3 to 15 MtCO2 for the first half of 2018. McCook (2018) estimated 63 MtCO2 per August 2018. Foteinis (2018) estimated the combined footprint of Bitcoin and Ethereum to be 43.9 MtCO2. One study even claimed that Bitcoin mining could cause emissions incompatible with the goal of the Paris Agreement to limit global warming to below +2 °C (Mora et al., 2018). These numbers become even more impressive given that the actual use of the Bitcoin network has remained limited. Over the course of 2019, the network processed 120 million transactions (Blockchain, 2020), while traditional payment service providers processed about 539 billion transactions (Capgemini, 2019). Dividing emissions estimates by the number of transactions yields a carbon footprint in the range between 233.4 and 363.5 kg of CO2 per Bitcoin transaction (de Vries, 2019). It is noteworthy that such annual estimates as depicted in Fig. 1 are typically based on results at a certain day assuming those daily conditions persisted for a year to facilitate comparisons with other emitting activities or national emissions on country level.

Over time, Bitcoin miners have turned to increasingly specialized hardware equipment with higher computing power (Bedford Taylor, 2017). Whereas miners initially used central processing units (CPUs) to find PoWs, they quickly realized that graphic processing units (GPUs) were better equipped for the task. In 2013, application-specific integrated circuits (ASICs) entered mining and quickly replaced GPUs as the standard hardware. As implied by the name, ASICs perform one specific task: finding the required proofs at optimal efficiency. In fact, ASICs are so specialized that they only fit one mining algorithm. Bitcoin ASIC-based mining devices cannot be used to mine any alternative digital currency. This hyper-specialization of devices also implies that miners rapidly cycle through vast amounts of increasingly powerful mining devices.

In this study, we demonstrate a methodology for estimating Bitcoin's e-waste. Firstly, we develop a framework to assess the current state of the network's e-waste generation. Secondly, we utilize the initial public offering (IPO) filing of a major hardware manufacturer to calibrate our framework. Lastly, we discuss strategies to mitigate the e-waste challenge of Bitcoin and the implications of the results in relation to the sustainability of digital currencies.

With our results, we aim to inform and broaden the debate on the environmental costs of cryptocurrencies. This debate should become more inclusive of externalities beyond the energy consumption of cryptocurrency mining devices and cryptocurrencies besides Bitcoin (Gallersdörfer et al., 2020). As Bitcoin continues to dominate the cryptocurrency market – with a market share of almost 70% per the start of 2021 (CoinMarketCap, 2021) – this study focuses solely on Bitcoin. Our results may serve as a reference point for emerging cryptocurrencies beyond Bitcoin for how much e-waste they can potentially generate. Our results may also help stakeholders better understand and mitigate the environmental impacts of digital currencies.

Generally, assessing e-waste with accurate estimates is difficult due to a lack of high-quality data (Wang et al., 2013). Most common estimation methods collect data through industry visits, surveys, and sales reports (Islam and Huda, 2019). The widely cited Global e-waste Monitor, a collaborative effort formed by the United Nations University, evaluates production, sales, and trade data along with appliance characteristics and expert knowledge to calculate e-waste (Forti et al., 2020). Similar inputs cannot easily be collected from the Bitcoin mining industry. Bitmain is the largest manufacturer of Bitcoin mining devices, with an estimated market share of 76% (Stoll et al., 2019); it only publicly disclosed sales information once before its planned IPO in 2018. As the IPO was canceled, there has been no need for Bitmain to continue disclosing its sales, and there is currently no reason to assume it will do so again in the (near) future. Likewise, obtaining reliable and representative survey responses from the industry may prove increasingly challenging because of a growing number of illegal facilities. The Bitcoin mining activities in Iran represent a growing percentage of all Bitcoin mining activities. The country powered almost 4% of all Bitcoin mining activities in April 2020 (University of Cambridge, 2020a). This share could amount up to 17% per May 2021. The annual energy consumption of Bitcoin miners in Iran amounts to 20 terawatt-hours (TWh; Tassev, 2021), while Bitcoin miners globally consume around 117 TWh annually as of May 2021 (Digiconomist, 2021). Within Iran, it has been suggested that more than 86% of the electricity used to power Bitcoin mining is obtained illegally (Tassev, 2021). This may introduce serious sample selection bias, specifically a survivorship bias, since any survey will oversample the licensed miners. This trend in illegal mining operations may be amplified by aggressive policies towards Bitcoin mining in other countries. China, for instance, is estimated to house most of the Bitcoin mining network (University of Cambridge, 2020a), but regulators in China's Inner Mongolia region have already moved to ban Bitcoin mining over environmental concerns (Barrett, 2021), and other provinces are taking similar actions (Reuters, 2021).

Despite these challenges, the Bitcoin mining industry may offer a unique opportunity to obtain a real-time estimate of the network's e-waste generation, because we can observe live data that allows us to estimate both the amount of active equipment in the network and the lifespan of devices used. Unlike other industries, developments in the total amount of active equipment can easily be observed as we can estimate the total amount of computations all active mining devices generate at any given moment. While a granular breakdown of which devices operate in the network is not immediately available, we can use public information on the characteristics of available devices to determine the likely amount of active equipment.

Application-specific integrated circuit chips like those used for Bitcoin mining devices are designed such that they are hardwired to perform a single repeated function. This makes them far more efficient at their specific task than the general-purpose chips used in CPUs and GPUs. In performance comparisons, it was found that an ASIC Cloud could perform 6270 times more operations per second on Bitcoin than a CPU Cloud and 1057 times more than a GPU Cloud (Khazraee et al., 2017). To understand how this has made CPUs and GPUs obsolete for mining Bitcoin, the following example is useful. In this example, all miners are competing for the same reward, and the network's protocol self-adjusts the difficulty of finding a valid proof to keep block production time constant; this means that the rewards do not grow as more mining devices enter the network. Instead, as the chance of mining a new block depends on the proportional share of computational power in the network, any increase in the total computational power of the network will marginally dilute the share of every individual device and thus reduce individual earning capacity (de Vries, 2019). As mining devices primarily require electricity to operate, miners can only obtain a competitive advantage by increasing their efficiency (i.e. using less energy per unit of computational power). This dynamic has resulted in a race to develop and deploy more efficient mining hardware, which causes the earning capacity of individual devices to decline. Because their operating costs stay the same, older and/or inefficient devices will be forced to leave the network once they operate at a loss. As CPUs and GPUs simply are not cost-effective at mining Bitcoin, they have been rendered obsolete for this purpose.

A similar dynamic ultimately determines the fate of ASIC-based mining devices as advances in ASIC chip efficiency result in more powerful devices that eventually crowd out older, less efficient technology (Fig. 2). Because the technical lifetime of ASIC mining devices typically exceeds the period of time during which the device can perform its task profitably (McCook, 2018), the moment they become unprofitable determines their lifespan and the point at which they become electronic waste. The fact that ASIC chips are single-purpose and not customizable prevents them from being repurposed for another task or even another type of cryptocurrency mining algorithm.

The rate at which these devices become obsolete has not been examined in detail. De Vries 2019) assumed that mining equipment becomes obsolete every 1.5 years based on a small selection of devices. In general, advances in ASIC-based device efficiency have historically outpaced Koomey's law, which describes the efficiency improvements of computing and shows that computations per unit of energy consumed double every 1.57 years (Koomey et al., 2011). Fig. 3 compares the expected efficiency gains in Bitcoin ASIC-based mining devices in relation to Koomey's law since 2014 against the actual efficiency of mining devices (see Supplemental Data: Sheet 3). The comparison shows that the speed of efficiency improvements of Bitcoin ASIC mining devices largely exceeded expectations in terms of required energy input per hash, measured in Gigahash per Joule (GH/J). These rapid improvements mean that older devices quickly lose their competitive edge, putting them at risk of being displaced by newer device types.

While it is a downside that ASIC-based mining devices cannot be repurposed after becoming obsolete for their single purpose, this property, along with our ability to assess the profitability of any device type in real-time using public data, provides us with another unique opportunity to estimate the lifespan of these devices. Their end of use and end of life is explicitly marked by the moment the equipment becomes unprofitable at mining. At this point, they will be disposed of and become e-waste. What happens to these machines depends on the respective location, as manufacturers like Bitmain offer no recycling programs. China has historically housed most of the Bitcoin network but formally collects only 16% of all e-waste generated. Other destinations such as Iran, Kazakhstan, and Malaysia perform even worse. None of these countries has a comprehensive e-waste regulation. In middle- and low-income countries, e-waste is mostly handled by the informal sector, which is known to cause severe damage to both the environment and human health (Forti et al., 2020). Given Bitcoin's substantial footprint in middle- and low-income countries, Bitcoin mining devices are likely to end up in this informal sector as well.

Section snippets

Materials and methods

To gauge the lifetime of Bitcoin mining devices, we used data from the Cambridge Center for Alternative Finance (University of Cambridge, 2021), which keeps track of available device types (along with different iterations of the same device). We combined this with publicly available product specifications that reveal computational power, power efficiencies, and equipment weight of a given device. We used this data to evaluate the duration of profitable operation per mining device, assuming that

Results

To gauge the e-waste generation of Bitcoin, we firstly determined the lifetime of Bitcoin mining devices based on their ability to operate profitably. Secondly, we used this information to derive the amount of active equipment in the Bitcoin network over time since July 2014 as well as the resulting e-waste.

Conclusion

Bitcoins’ growing energy consumption and carbon footprint have received wide attention from the general public and kickstarted academic debate for the past few years. However, most research has overlooked the environmental impact of the usage and disposal of raw materials in the highly specialized mining equipment responsible for the energy consumption in the first place. We show in this study that the lifespan of Bitcoin mining devices remains limited to just 1.29 years. As a result, we

Discussion

Here we will discuss how this growing need for hardware has implications beyond e-waste generation and what might be done to mitigate the concerns.

Data availability

All data used in this analysis are publicly available online under the noted sources.

Funding

The authors did not receive support from any organization for the submitted work.

CRediT authorship contribution statement

Alex de Vries: Conceptualization, Data curation, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing. Christian Stoll: Conceptualization, Investigation, Methodology, Validation, Writing – review & editing.

Declaration of Competing Interest

The authors have no conflicts of interest to declare that are relevant to the content of this article.

References (49)

  • M. Bedford Taylor

    The evolution of Bitcoin hardware

    Computer (Long Beach Calif)

    (2017)
  • BitInfoCharts, 2020. Monero Hashrate historical chart. URL...
  • Bitmain, 2019. S9 Specifications. URL...
  • Blandin, A., Pieters, G.C., Wu, Y., Eisermann, T., Dek, A., Taylor, S., Njoki, D., 2020. 3rd Global cryptoasset...
  • Blockchain, 2020. Confirmed Transactions Per Day. URL...
  • Canaan Inc., 2019. Amendment no.1 to form F-1 registration...
  • Capgemini, 2019. World Payments Report 2019. URL...
  • H. Cho

    ASIC-resistance of multi-hash proof-of-work mechanisms for blockchain consensus protocols

    IEEE Access

    (2018)
  • Coin Metrics, 2020. Coin Metrics’ State of the Network: Issue 51 [WWW Document]. URL...
  • Percentage of Total Market Capitalization (Dominance)

    (2021)
  • CoinTelegraph, 2021. Bitmain ceases Bitcoin miner sales to aid second-hand sellers following China ban [WWW Document]....
  • Colvin, G., 2018. EIP-1057. GitHub. URL...
  • Cryptoslate, 2020. PoS Coins. URL https://cryptoslate.com/cryptos/proof-of-stake/ (accessed...
  • A. de Vries

    Bitcoin Boom: What Rising Prices Mean For the Network's Energy Consumption

    (2021)
  • Cited by (82)

    • The changing dynamics of crypto mining and environmental impact

      2024, International Review of Economics and Finance
    View all citing articles on Scopus
    View full text