AI's Massive Energy Appetite Raises Concerns About Environmental Sustainability
-
AI's computing power requires a lot of electricity, and its energy use could reach 134 TWh per year by 2027 - comparable to some countries' consumption.
-
Data centers' energy use has increased 6% from 2010-2018, but AI's growth could cause a surge. If Google used AI like ChatGPT it could use 29.2 TWh annually.
-
NVIDIA will deliver 100,000 AI servers this year, using 5.7-8.9 TWh. By 2027 with 1.5 million servers it could reach 85-134 TWh - significant for data centers.
-
AI has issues like privacy, biases, and hallucinations. Environmental sustainability is another concern as its energy use grows.
-
Expert argues AI isn't a "miracle cure" and its energy impact needs careful review of where and when to use it.