Satya Nadella, CEO of Microsoft, says he spends his weekends studying how startups build products to help the tech giant regain the speed and agility it lost as it grew. Speaking in an interview on the "MD MEETS" podcast hosted by Mathias Döpfner, CEO of Axel Springer, Nadella revealed he’s actively trying to relearn the fast-paced, collaborative culture of young companies. He described how, in startups, product development teams — including scientists, engineers, and infrastructure specialists — often work together at a single table, enabling rapid decision-making and instant iteration. At Microsoft, he noted, those roles are managed by three separate divisional heads, creating layers of bureaucracy that slow progress. Nadella’s comments come amid a broader trend across Silicon Valley, where major tech companies like Meta, Google, and Amazon are streamlining management structures to accelerate innovation. He sees Microsoft’s size as a "massive disadvantage" in the race to build AI at startup speed. To stay relevant in the AI era, Nadella stressed the need to unlearn the habits that once defined success. "The most important skill set for long-term relevance is — how do you be a learn-it-all and not a know-it-all," he said. He added that leaders must be willing to shed the "know-it-all" mindset and embrace continuous learning, especially as AI transforms industries. Empathy and emotional intelligence, he said, are now critical leadership traits in a world where technology and human experience are increasingly intertwined. Internally, Nadella has already begun reshaping Microsoft’s culture. A leaked organizational chart reviewed by Business Insider showed he now has 16 direct reports — a group of carefully selected executives tasked with breaking down silos and driving the company’s AI transformation. Nadella also outlined four essential steps companies must take to succeed with AI, warning that most corporate AI initiatives fail because they’re treated like traditional IT upgrades — a flawed approach he called "going to fail by definition." The four steps: completely rethink workflows, adopt modern AI tools, train employees to use them effectively, and ensure company data isn’t trapped in outdated legacy systems. Only organizations that rebuild these foundations, he said, will achieve real value from AI — and only leaders willing to unlearn their past successes will be able to lead the change.