Policy assessments for the carbon emission flows and sustainability of Bitcoin blockchain operation in China

The study addresses the escalating energy consumption and carbon emissions associated with Bitcoin’s Proof-of-Work consensus and evaluates their implications for China’s sustainability targets under the Paris Agreement. The authors aim to quantify current and future carbon emission patterns of Bitcoin blockchain operations in China and assess the effectiveness of different policy interventions. They develop a system dynamics-based Bitcoin Blockchain Carbon Emission (BBCE) model that captures feedback loops across three subsystems—mining and transactions, energy consumption, and carbon emissions—to simulate scenarios and inform policy design.

The paper situates its work within prior research indicating that Bitcoin’s PoW has driven rapid hardware evolution (CPU to GPU to ASIC) and an arms race that boosts energy demand and capital expenditures. Earlier estimates suggest Bitcoin’s annual energy use can match that of small to medium nations (e.g., Denmark, Ireland) and that between Jan 2016 and Jun 2018 up to 13 million metric tons of CO2 may be attributable to Bitcoin. China accounts for over 75% of the Bitcoin network’s hashing power due to proximity to hardware manufacturers and low-cost electricity. System dynamics modeling has been applied in climate and industrial contexts to assess carbon flows and policy impacts, offering advantages in capturing endogenous dynamics and evaluating disequilibrium scenarios. These strands motivate the BBCE approach to assess policies such as market access standards, site regulation, and carbon taxation for Bitcoin mining in China.

The authors construct the BBCE (Bitcoin Blockchain Carbon Emission) system dynamics model with three interconnected subsystems: (1) Bitcoin mining and transaction, (2) energy consumption, and (3) carbon emissions. Key assumptions include: (1) electricity for mining is coal-based or hydro-based; (2) long-term Bitcoin price dynamics are driven by block reward halving cycles (price extremely volatile in reality); (3) miners exit or relocate if mining becomes unprofitable in China; (4) policy levers (market access, site regulation, carbon tax) align with national emission management; (5) miners maintain full investment intensity while operating. - Structural elements and equations: Investment intensity increases over time and with the proportion of Chinese miners; miner cumulative profits integrate profit rate minus investment intensity; network hash rate depends on investment intensity; block size increases with time; the proportion of Chinese miners declines if cumulative profits turn negative; mining efficiency rises with investment and tighter market access standards; mining power equals hash rate times efficiency; network energy consumption equals mining power times power usage effectiveness (PUE); total carbon emission integrates monthly flows, summing coal-based and hydro-based emissions; carbon emission per GDP equals emissions divided by modeled GDP (derived from miners’ profit rate and total cost); a carbon tax is applied and doubled if carbon emission per GDP exceeds 2 kg/USD (initial 1% tax based on World Bank data). - Scenarios (2014–2030, with focus on 2021–2029 annualized outcomes): Benchmark (BM) with market access 100%, 40% of miners in coal regions, carbon tax threshold at 2; Market Access (MA) doubles efficiency standard (selects out low-efficiency miners); Site Regulation (SR) shifts miners from coal regions so only 20% remain in coal; Carbon Tax (CT) doubles the punitive carbon tax (to 4 when triggered). Scenario parameterization summarized in Table 1. - Data and parameterization: Model implemented in Vensim PLE 8.2.1. Time-series inputs from btc.com (hash rate, block size, transaction fee, difficulty). Carbon intensities from Cheng et al. Energy cost and carbon taxation from World Bank. Miner site proportions from btc.com pool regional stats. Monthly historical data (Jan 2014–Jan 2020) used for parameter regression (Stata 14.1). Simulation period: Jan 2014–Jan 2030. Initial/static parameters in Supplementary materials. - Validation: Structural suitability, reality/statistical tests, and sensitivity analysis confirm behavioral robustness and consistency with observed data. Minor parameter variations do not change policy ranking outcomes.

- Under Benchmark (minimal intervention), annual energy consumption in China’s Bitcoin operations is projected to peak at 296.59 TWh in 2024, with corresponding carbon emissions of 130.50 million metric tons (Mt). This energy use exceeds that of Italy and Saudi Arabia (2016 levels) and ranks 12th globally; emissions surpass those of the Czech Republic and Qatar (2016) and would place Bitcoin mining among the top 10 emitters across 182 Chinese cities and 42 industrial sectors, accounting for approximately 5.41% of emissions from China’s electricity generation. Peak carbon emission per GDP reaches 10.77 kg/USD. - Scenario outcomes: • CT (Carbon Tax): Peak energy demand decreases to 217.37 TWh; carbon emissions reduced relative to BM but with limited effectiveness, primarily until about July 2024. • MA (Market Access): Total energy consumption increases (peak 350.11 TWh in 2024), and carbon emissions rise, reaching 140.71 Mt in 2025, indicating incentive effects that prolong profitability for remaining efficient miners. • SR (Site Regulation): Energy consumption peaks at 319.80 TWh (2025) yet carbon emissions and carbon intensity decline substantially due to relocation toward hydro-rich regions; peak carbon emission per GDP drops to about 6 kg/USD (roughly half of BM). - Additional dynamics: In BM, miners’ profit rate drops to zero in April 2024, triggering gradual relocation; network hash rate reaches about 1775 EH/s; miner total cost peaks around USD 1,268 million. Despite eventual profit pressures, energy consumption remains positive until 2030 due to sunk costs and gradual exits. - Overall, policies that change the energy mix (SR) outperform punitive taxes (CT) and market access restrictions (MA) in curbing emissions intensity and total emissions from Bitcoin mining in China.

The findings show that Bitcoin’s PoW mechanism creates positive feedback loops—hardware arms races and rising hash rates—that drive high energy use and emissions. The BBCE simulations indicate that without intervention, Bitcoin mining in China becomes a significant impediment to national emission reduction goals under the Paris Agreement, with high carbon intensity per GDP and substantial contributions to national emissions. Policy comparisons reveal that intuitive punitive measures, notably carbon taxes, have limited temporal and quantitative effectiveness, while market access tightening can inadvertently increase emissions via incentive effects among efficient survivors. In contrast, site regulation that shifts mining from coal-dominated to hydro-rich regions effectively reduces carbon intensity and overall emissions by altering the energy consumption structure. The study underscores the need for sector-specific accounting for Bitcoin’s emissions to enable targeted regulation, and it highlights the broader implication that PoW-based blockchain operations are environmentally intensive, potentially undermining sustainability unless protocols or operational policies evolve.

This paper develops a system dynamics-based BBCE model to quantify Bitcoin mining’s energy use and carbon emissions in China and to evaluate policy effectiveness. It shows that, absent intervention, energy use and emissions peak mid-decade at levels comparable to or exceeding those of several countries and major Chinese sectors. Scenario analysis demonstrates that site regulation, which shifts mining toward low-carbon electricity, is more effective than punitive carbon taxes or market access restrictions in reducing carbon intensity and total emissions. The work contributes a modeling framework for emergent, unaccounted sectors and offers policy guidance emphasizing energy mix transformation over solely punitive measures. Future research should refine price modeling beyond halving-linked linearity, incorporate relocation costs, consider evolving national electricity mixes and policy changes, and explore alternative, less energy-intensive consensus mechanisms to ensure blockchain’s sustainable deployment.

Key limitations include: (1) Bitcoin price dynamics are simplified to follow halving-linked linear increases for long-term modeling, despite real-world volatility and investor-driven factors; (2) the SR scenario assumes zero relocation costs for miners, whereas actual costs (e.g., transportation, infrastructure) may reduce effectiveness; (3) except for SR, the model largely holds China’s electricity mix constant, not fully reflecting potential shifts toward renewables and coal pricing changes; (4) projections are subject to unforeseeable technological, market, and policy changes that could alter mining behavior and emissions; (5) Bitcoin operations are not an independently accounted sector in national statistics, complicating validation and policy targeting.