Satya Nadella, the CEO, has spearheaded artificial intelligence initiatives for years, thanks to his and the innovative work of his AI chief, Mustafa Suleyman—especially regarding . Yet Nadella hasn’t frequently addressed the concerns that shook Wall Street during most of the latter half of 2025: whether AI is a speculative bubble.
At the World Economic Forum’s annual meeting in Davos, Switzerland, Nadella sat down with the forum’s interim co-chair, CEO Larry Fink, to explain that if AI growth comes only from investment, that could indicate a bubble. “A clear sign it’s a bubble would be if all we discuss are tech companies,” Nadella stated. “If our only focus is the technology aspect, then it’s purely a supply-side phenomenon.”
But Nadella proposes a solution to this productivity problem: he urges business leaders to rethink knowledge work by adjusting workflows to align with AI’s structural design. “As leaders, our mindset should be this: we need to consider modifying the work—specifically the workflow—alongside the technology,” he said.
Growing pains
As Nadella noted, this shift isn’t entirely new; he drew a parallel between today’s era and the 1980s, when computing transformed workplaces, unlocked fresh growth and productivity opportunities, and created a new category of employees. “We developed the whole concept of knowledge work, where people began using computers to enhance what we aimed to accomplish with software,” he explained. “I believe the same will occur with AI.”
Nadella contends that AI completely reverses how information circulates within a company, replacing slow, hierarchical procedures with a model that compels leaders to reevaluate their organizational frameworks. “We have organizations, departments, and specializations, and information usually filters upward,” Nadella said. “But AI actually flattens the entire flow of information. Once that happens, you must restructure your organization.”
This transition might be more difficult for some 500 firms, as structural changes often bring uncomfortable growing pains. Nadella observes that smaller, leaner companies can adopt AI more easily since their organizational structures are newer and more flexible. In contrast, larger corporations may need more time to implement new workflows.
Even though AI is widely adopted, the 29th edition of revealed that only 10% to 12% of companies saw revenue or cost benefits from the technology, while 56% gained no value from it. This builds on a more negative earlier finding about AI returns: 95% of generative AI pilot projects were not succeeding.
PwC’s global chairman, Mohamed Kande, talked to ’s Diane Brady at Davos about the survey result that many CEOs are cautious and lack confidence in the current phase of AI adoption. “AI moves so quickly that people forgot technology adoption requires going back to basics,” he said. The survey found that companies benefiting from AI are “laying the groundwork.” Success depends more on execution than the technology itself, he emphasized, and strong management and leadership will be crucial in the future.
“For big organizations,” Nadella told Fink, “there’s a core challenge: If your pace of change doesn’t keep up with what’s feasible, you’ll be outperformed by a smaller player that can scale quickly using these tools.”
Newcomers have the edge of starting from scratch and building workflows around AI’s capabilities, whereas larger companies will have to deal with AI’s flattening impact on entire departments and specialized roles.
It’s important to note that Nadella acknowledges large organizations still hold an advantage—particularly in terms of relationships, data, and expertise. But he insists these firms must learn to leverage those resources to adjust their management styles; otherwise, this could become a significant barrier.
“The key takeaway is: if you don’t apply those resources through a new production model, you’ll end up stuck,” he concluded.