An IMF blog of 15 August 2024, written by Shafik Hebous and Nate Vernon-Lin, indicated that tax policy can be used to restrict the surge in carbon emissions resulting from artificial intelligence (AI) and crypto.

Owing to the amount of electricity used by the high-powered equipment employed in mining crypto assets, the authors point out that each Bitcoin transaction requires roughly the same amount of electricity that would be consumed by the average person in Ghana or Pakistan in three years. In 2022, crypto mining and data centers accounted for 2% of global electricity demand. That share is likely to climb to 3.5% in three years, according to IMF estimates based on projections from the International Energy Agency (IEA). That is approximately the same share of global electricity demand as the current consumption of Japan, which is the world’s fifth largest electricity user.

The impact of AI and crypto on the climate has become a cause for concern. A recent IMF working paper noted that crypto mining could generate 0.7% of global carbon dioxide emissions by 2027. If this estimate is also extended to data centers (based on IEA estimates), the combined carbon emissions of AI and crypto could reach 450 million tons by 2027, which would amount to 1.2% of the world total.

The authors indicate that one way to restrict the carbon emissions from these sources is through the tax system. IMF estimates indicate that a direct tax of USD 0.047 per kilowatt hour could encourage the crypto mining industry to limit its emissions and bring them generally in line with global targets. Taking into account the impact of air pollution on health, the required tax rate would rise to USD 0.089, representing an 85% percent rise in the average electricity price for data miners. This levy could increase the annual revenue of governments around the world to USD 5.2 billion and reduce annual emissions by 100 million tons.

In the case of data centers, a tax on their electricity use would need to be set at USD 0.032 per kilowatt hour, or USD 0.052 including air pollution costs. A tax imposed at this rate could raise up to USD 18 billion annually.

An opposing view would be that current government policies involving incentives and exemptions for AI could result in applications that lead to smarter and more efficient power use, helping to ease electricity demand. However, if governments design the right policies, they could continue to incentivize developing AI applications while also introducing taxes that help to address the environmental damage.

The authors note that a broad carbon price coordinated across countries would be the best way to restrict emissions, reducing fossil-fuel consumption and improving energy efficiency. However, in the absence of a global carbon price, targeted measures could be used to encourage crypto miners and data centers to employ more energy-efficient equipment. Electricity taxes combined with credits for zero-emission, bilateral power purchase agreements and potentially renewable energy certificates could be effective. Cross-border coordination is also important, to prevent relocation by polluters to jurisdictions with lower standards.