Forget Bitcoin — AI Could Be the Next Energy Consumption Boogeyman
For years, the mainstream media has demonized Bitcoin over its energy consumption. The rise in artificial intelligence (AI) and the power-hungry processors that drive it may push aside the world’s most popular digital currency.
The pivot of Bitcoin mining to renewable power sources may give the network some reprieve over its energy consumption. Moreover, AI is taking the spotlight, and it is also a big consumer of energy, according to recent research.
AI Energy Usage in the Spotlight
Artificial intelligence is expected to have a huge impact on society, just like the internet. However, it will require a massive amount of additional energy to do so.
On Oct. 1, Bitcoin pioneer Nic Carter commented on the rise of AI and the power-hungry hardware needed to run it.
“Everything they said about Bitcoin mining they will say about AI – especially if “unregulated” open source models emerge that they can’t control and regulate,”
The comments come in response to research from energy expert Brian Gitt. “New AI chips use an insane amount of electricity,” he said before adding:
“A single Nvidia GPU chip consumes about the same amount of energy as a typical US home.”
Nvidia GPUs alone will consume more than the entire planned US data center capacity expansions over the next decade, he reported.
Researchers at the University of Washington revealed that hundreds of millions of queries on ChatGPT can cost around 1 gigawatt-hour a day. Furthermore, this is the equivalent of energy consumed by 33,000 US households, according to reports.
Moreover, the data centers that power cloud computing are becoming more reliant on power-hungry graphics processing units (GPUs) to run AI models.
Data center energy consumption. Source: X/@BrianGitt
A professor of electrical and computer engineering, Sajjad Moazeni, said:
“The energy consumption of something like ChatGPT inquiry compared to some inquiry on your email, for example, is going to be probably 10 to 100 times more power hungry,”
Energy Crisis Predicted
Additionally, Arijit Sengupta, founder and CEO of Aible, an enterprise AI solution company, added:
“The world is actually headed for a really bad energy crisis because of AI unless we fix a few things.”
Energy usage for data centers grew 25% annually from 2015 to 2021, far outpacing the growth in renewable energy deployment.
Furthermore, researchers are looking at ways data centers can shift computation to times when more renewable energy is available on the grid.
In September, it was reported that Microsoft was looking at next-generation nuclear reactors to power its data centers and AI.
Bitcoin may no longer be the energy bad boy. Last month, Bloomberg analyst Jamie Coutts reported that more than 50% of mining is now done using sustainable energy.