Microsoft CEO Satya Nadella has been leading the charge on artificial intelligence for years, owing to his long alliance with OpenAI’s Sam Altman and the groundbreaking work from his own AI CEO, Mustafa Suleyman, particularly with the Copilot tool. But Nadella has not spoken often about the fears that rattled Wall Street for much of the back half of 2025: whether AI is a bubble.

At the World Economic Forum’s annual meeting in Davos, Switzerland, Nadella sat down for a conversation with the forum’s interim co-chair, BlackRock CEO Larry Fink, explaining that if AI growth spawns solely from investment, then that could be a sign of a bubble. “A telltale sign of if it’s a bubble would be if all we are talking about are the tech firms,” Nadella said. “If all we talk about is what’s happening to the technology side then it’s just purely supply side.”

However, Nadella offers a fix to that productivity dilemma, calling on business leaders to adopt a new approach to knowledge work by shifting workflows to match the structural design of AI. “The mindset we as leaders should have is, we need to think about changing the work—the workflow—with the technology.”

Growing pains

This change is not wholly unprecedented, as Nadella pointed out, comparing the current moment to that of the 1980s, when computing revolutionized the workplace and opened up new opportunities for growth and productivity and created a new class of workers. “We invented this entire class of thing called knowledge work, where people started really using computers to amplify what we were trying to achieve using software,” he said. “I think in the context of AI, that same thing is going to happen.”

Nadella argues that AI creates a “complete inversion” of how information moves through a business, replacing slow, hierarchical processes with a view that forces leaders to rethink their organizational structures. “We have an organization, we have departments, we have these specializations, and the information trickles up,” Nadella said. “No, no, it’s actually—it flattens the entire information flow. So once you start having that, you have to redesign structurally.”

That shift may be harder for some Fortune 500 companies as structural changes could be accompanied by uncomfortable growing pains. Nadella says that leaner companies will be able to more easily adopt AI because their organizational structures are fresher and more malleable. On the other hand, large companies could take time to adopt new workflows.

Despite widespread adoption of AI, the 29th edition of PwC’s Global CEO Survey found that only 10% to 12% of companies reported seeing benefits from the technology on the revenue or cost side, while 56% reported getting nothing out of it. It follows up on an even more pessimistic finding about AI returns from August 2025: that 95% of generative AI pilots were failing.

PwC global chairman Mohamed Kande spoke to Fortune’s Diane Brady at Davos about the finding that many CEOs are cautious and lack confidence at this stage of the AI adoption cycle. “Somehow AI moves so fast … that people forgot that the adoption of technology, you have to go to the basics,” he explained, with the survey finding that the companies seeing benefits from AI are “putting the foundations in place.” It’s about execution more than it is about technology, he argued, and good management and leadership are really going to matter going forward.

“For large organizations,” Nadella told Fink, “there’s a fundamental challenge: Unless and until your rate of change keeps up with what is possible, you’re going to get schooled by someone small being able to achieve scale because of these tools.”

New entrants have the advantage of “starting fresh” and constructing workflows around AI capabilities, while larger firms will have to contend with the flattening effect AI has on entire departments and specializations.

To be sure, Nadella says that large organizations have kept an upper hand, especially when it comes to relationships, data, and know-how. However, he maintains that firms must understand how to use those resources to their advantage to change management style, which could otherwise pose a major roadblock.

“The bottom line is, if you don’t translate that with a new production function, then you really will be stuck,” he said.