Is It Time to Tax Data?

The explosion of data is an environmental hazard, and AI will make it worse

In 2025 the world will produce 180 zettabytes of data. That’s one byte for every star in the known universe, and a threefold increase from just five years ago. Moreover, the trendline is still pointing up. By 2028, it’s estimated the total amount of data we produce will more than double again.

It takes a lot of energy to generate, process, and store all that data. In 2022 data centers used 3 percent of all electricity generated in the United States. By 2030, that will more than double to 8 percent. This data explosion is largely due to the adoption of AI, which is only getting started. And AI is very energy hungry. Depending on who you ask, those AI Overviews provided by Google’s Gemini take anywhere from 10 to 30 times as much energy to process as a normal Google query.

The environmental impact of creating and storing all this data will be staggering. One petabyte of data takes 300,000 kilowatt hours of electricity per year to store. To generate that much electricity costs companies on average $240,000, and produces 240 metric tons of carbon dioxide per year. And there are a million petabytes in just one zettabyte.

The obvious question, then, is what the emergence of AI is going to mean for our efforts to fight global warming? Can we possibly get carbon emissions under control while also developing and deploying AI tools with the speed and breadth that businesses seem to want?

That may be a tall order. Already Google and Microsoft are second-guessing their pledges to reach carbon-neutrality by 2030. Other companies, with less commitment to sustainable innovation, probably won’t even try. So what to do?