Artificial intelligence (AI) chips have come under scrutiny from energy and tech experts for their significant electricity consumption. Energy entrepreneur Brian Gitt recently took to X (formerly Twitter) to report the rising energy consumption by AI chips. According to Gitt, a single Nvidia graphics processing unit (GPU) consumes approximately the same amount of energy as a typical house in the United States.
Citing data gathered by finance expert Duncan Simmons, Gitt told his followers on X that Nvidia GPUs are on track to consume more energy than the entire planned US data center capacity expansions over the next decade. In 2023, data centers will reportedly consume 50% more electricity than they did in 2020.
Responding to Gitt’s tweet, crypto influencer Nic Carter pointed out that Bitcoin mining was also subjected to the criticisms that AI is currently facing. According to Carter, the criticism for AI will continue given the unregulated open-source nature of AI models, which are hard to control or regulate by a centralized authority.
Bitcoin mining has been under scrutiny from environmentalists who have repeatedly highlighted the energy-intensive mining process. In late 2017, the World Economic Forum (WEF) published a report claiming that in 2020, Bitcoin would consume more power than the world did at the time. Similar criticisms were previously directed towards emerging technology like the internet.
Earlier this year, Duncan Simmons published a report that took a closer look at developers’ pivot from Bitcoin mining to high-performance computing and artificial intelligence. Citing recent trends, Simmons stated that Bitcoin miners transitioned to the high-performance computing market after BTC’s price took a hit. Bitcoin developers are reportedly well-positioned to contribute to the AI market, given their experience with large-scale and high-capacity operations.