June 8, 2024

AI vs Bitcoin Mining: Which Consumes More Energy?

Artificial intelligence (AI) and cryptocurrency mining are two of the most energy-intensive technologies of our time. As these technological giants continue to expand, a pressing question emerges: which one has a larger appetite for energy?

AI's Energy Consumption

AI technologies, especially large models like GPT, require significant amounts of energy to function. These models operate on extensive data centers filled with energy-hungry machines performing complex computations. Servers equipped with multiple GPUs (or TPUs, in the case of Google) are typically provided by Nvidia. Each AI server consumes about two kilowatts continuously, according to Shaolei Ren, a researcher from the University of California Riverside.

Scale of AI Usage:

  • Large Models: For a single large AI model like GPT, SemiAnalysis estimates that if ChatGPT were integrated into every Google search, it would necessitate over 500,000 of Nvidia’s A100 HGX servers.
  • Energy Usage: This would result in a daily electricity consumption of 80 gigawatt hours (GWh) and an annual consumption of 29.2 terawatt hours (TWh), according to Alex de Vries from the Vrije Universiteit Amsterdam.

Future Projections:

  • Growing Demand: With AI technologies like Google's Gemini and OpenAI's secret "Q*" project on the rise, the demand for energy is expected to skyrocket. de Vries projects that AI servers could consume between 85 to 134 TWh annually by 2027, comparable to the annual power usage of Norway or Sweden, and representing 0.5% of global power demand.

Environmental Concerns:

  • Carbon Footprint: Professor Martin Takáč of MBZUAI emphasizes the need for researchers to consider the environmental impact of AI operations. At MBZUAI, researchers account for the carbon footprint of AI models in their designs.

Crypto Mining's Energy Consumption

Crypto mining, particularly Bitcoin mining, is also known for its substantial energy usage. The University of Cambridge Centre for Alternative Finance (CCAF) provides some of the most accepted estimates on this front. According to the Cambridge Bitcoin Electricity Consumption Index (CBECI), Bitcoin mining consumes around 112.31 TWh annually, nearly equivalent to the annual electricity consumption of the Netherlands.

Mechanism of Consumption:

  • Proof of Work (PoW): Bitcoin's energy consumption stems from its PoW consensus mechanism, where miners compete to solve complex mathematical puzzles using computational power to add blocks to the blockchain.

Comparative Consumption:

  • Other Cryptocurrencies: Not all cryptocurrencies are as energy-intensive. Ethereum, for example, uses a Proof of Stake (PoS) mechanism, which consumes significantly less energy—around 7.08 GWh annually.

Water Consumption

Both AI and Bitcoin mining data centers have high water usage for cooling purposes.

AI Data Centers:

  • Water Usage: Microsoft's global water consumption rose by 34% from 2021 to 2022, reaching nearly 6.4 gigaliters (GL). Google reported a 20% increase in the same period.
  • Usage Per Inquiry: Ren’s research suggests that each time a user asks ChatGPT five to 50 questions, the chatbot uses about 500 milliliters of water. This translates to an estimated 4,200 to 6,600 GL of water usage by AI data centers globally by 2027.

Bitcoin Mining:

  • Water Usage: Bitcoin mining also demands significant water for cooling. It is estimated that Bitcoin consumed over 1,600 GL of water in 2021, potentially rising to more than 2,200 GL in 2023.

Environmental Incidents:

  • Lawsuits: Environmental groups in the US sued Greenidge Generation, a Bitcoin mining company, for allegedly discharging hot water into Seneca Lake, raising concerns about water temperature regulation. The lawsuit was later dismissed.

Renewable Energy and Future Solutions

Bitcoin Mining:

  • Green Energy: Approximately 70% of the energy used in Bitcoin mining comes from renewable sources, driven by miners’ efforts to generate their own power, as noted by Bijan Alizadeh of Phoenix Group.

AI Data Centers:

  • Fossil Fuels: The majority of AI data centers still rely on fossil fuels. Tech giants like Microsoft are exploring the use of small modular nuclear reactors to power their data centers, although this solution is seen as extreme by some experts like de Vries.

Regional Innovations:

  • Balancing the Grid: In the UAE, crypto mining facilities help balance the electricity grid during low demand periods, like winter, by utilizing excess energy for Bitcoin mining.

While both AI and Bitcoin mining have significant energy and water footprints, Bitcoin mining shows a higher dependence on renewable energy. Meanwhile, the tech industry continues to seek sustainable solutions to power its expansive data centers and mitigate environmental impacts.

📌 WEB

📌 TELEGRAM

📌 TELEGRAM CHAT