What Is a ‘Compute Tax’ and Why Is the Idea Gaining Traction?
The extent of AI’s economic impact is still up for debate while some are already pondering solutions to worst-case scenarios
The ripple effects of artificial intelligence on the economy are only expected to intensify, from job displacement to a shift in GDP from labor to capital. The unknowns are prompting some to revisit an old idea with a fresh face: a tax on AI processing, aka “compute.”
Sound familiar? Bill Gates proposed the idea of a “robot tax” back in 2017—well before ChatGPT and Claude Code were household names.
“Half a year ago, it was something you would hear about only in very select circles,” says Anton Korinek, an economics professor currently on leave from the University of Virginia Darden School of Business. “It really has become much more mainstream in the last three months.”
Advocates support an AI tax to make up for potentially massive job loss and to slow AI’s extreme growth rate.
“We’re at a point now where we need to try and preserve jobs,” says Andrew Yang, a former presidential candidate and a co-chair of the Forward Party. “AI is going to gut white-collar employment and then by the way, it’s going to proceed to truck driving and a lot of other very common jobs.” Yang has long supported a so-called universal basic income.
“If you look at the taxes that are being paid by the biggest AI companies, they’re nowhere near commensurate to the value that AI is going to end up both generating and soaking up,” he adds.
Last month, Texas billionaire John Arnold, who co-chairs the philanthropic organization Arnold Ventures, floated the compute-tax idea on X.
“Only way to limit coming AI backlash is to start shifting taxes from labor to compute,” he wrote. “The average voter needs to see salient benefits from AI.” (Arnold wasn’t available to comment.)
In 2021, Sam Altman wrote, “We need to design a system that embraces this technological future and taxes the assets that will make up most of the value in that world—companies and land—in order to fairly distribute some of the coming wealth.”
The OpenAI chief recently told The Atlantic CEO Nicholas Thompson that “I no longer believe in universal basic income as much as I once did.”
Economists predict that the debate will only grow ahead of the next election season. With conversations already bubbling up in response to public sentiment about AI and doom and gloom predictions abounding, here’s what to know.
Why tax AI?
Generally there are two reasons to tax something, says Korinek: raising money and reducing the use of something by making it more expensive. A compute tax could bring revenue from AI’s success or slow its possible disruption. “You hear people in both camps of debate,” he says.
Many people are concerned about massive job loss as a result of AI, the concentration of the market, AI’s consumption of resources, and the community impact of expanding data centers. Some see taxation as a way to offset the loss of labor taxes to fund social services.
Yang advocates spreading the newly generated wealth around.
“If I were to generate even tens of billions of dollars in new tax revenue and it just went into the federal government and no one saw a dime of that, it wouldn’t actually make people feel any better about it,” he says.
Others want to put the brakes on AI, on account of safety concerns and perceived existential risks. In this argument, the compute tax would be akin to a pollution tax.
Who would get taxed?
In one scenario, companies that operate data centers would be taxed. In another, corporations and other users would be taxed on their consumption of tokens, AI’s key unit of measurement.
“If you had a tax on electricity, you would tax those companies that are distributing electricity, or you could tax the users of electricity,” says Pascual Restrepo, an associate professor of economics at Yale University.
Whether the Amazons and Microsofts are taxed at the front end, or the businesses and customers on the receiving end have to pay, the cost of using AI would go up, he says.
A compute tax is a sensible policy lever to consider in order to slow down automation, says Simon Johnson, a Nobel laureate and professor of entrepreneurship at the MIT Sloan School of Management. The tax could make it less appealing to fire tens of thousands of workers and build data centers, he adds. As AI advances, there should also be new things for humans to do.
“You want to create new tasks, you want to boost the demand for labor and that’s where the tech sector is kind of falling down,” he says.
The antitax argument
Korinek, the UVA professor, says that for now the amount spent on AI computation isn’t so big that a tax on it would provide a windfall. “It would raise a little bit of money but it would not really make a significant impact,” he says.
Restrepo notes that in addition to automating office work, the technology is already being used in drug discovery, weather forecasting, epidemiological models, fraud detection and many other important initiatives.
“Why do you want to increase the cost of all of that?” he says. Taxing compute could also just wind up pushing the U.S. industry abroad. “If you ask economists, most of them think this is unnecessary or too blunt,” he adds.
Any alternatives?
Given that AI is expected to transform society, many economists believe policy changes make sense. But some propose issue-focused solutions rather than a blanket tax. Restrepo says the U.S. already has a rich tax system that can tax corporate income, which might suffice even as AI progresses.
For AI safety issues, regulations might be better than taxes. And with labor, other policies could be adjusted, says Erik Brynjolfsson, director of the Stanford Digital Economy Lab.
In the current U.S. tax structure, a company with 1,000 workers pays more total taxes, including payroll taxes, than a company that makes the same amount of money using 1,000 machines, he says.
“We’re discouraging you from doing the one that amplifies humans and encouraging the one that replaces humans,” says
Brynjolfsson. “I don’t know if that was the intent, but that’s the effect.”
While a realignment of labor and capital taxes could help, he says, the broader solution is for AI to augment work instead of replacing it, and for the labor market to become more dynamic with portable benefits and job retraining.
“The compute tax kind of takes aim at the wrong thing,” says Brynjolfsson.