Satya Nadella Says Engineers Should Learn AI by Building, Not Chasing Hype
Doing beats discussing
Satya Nadella's advice to software engineers is blunt: stop reading about AI and start building with it. On a recent episode of The Pragmatic Engineer podcast, the Microsoft CEO argued that hands-on practice beats following the conversation. Coming from a chief executive, that lands differently. Executives usually talk about AI in terms of strategy, productivity and customers, so hearing one tell engineers to get their hands dirty is a shift in emphasis. Remember whose company makes many of these tools. That doesn't make the advice wrong, but it's fair to ask who benefits when more people start building.
For engineers just starting out, his prescription is concrete. Find a team that's shipping with the technology. Nadella told host Gergely Orosz, a software engineer and author, that the value is in the doing. "It's just not about the hype cycle and that constant need to talk," he said. "But it's do more, learn more and hands-on experience — experience it firsthand."
That distinction matters more than it sounds. Plenty of people follow AI news closely without ever shipping anything that depends on it, and the gap between knowing the vocabulary and knowing where a model fails in production is wide. Nadella's point is that the second kind of knowledge comes from work, not commentary. A lot of commentary goes stale quickly, and a working habit doesn't. If you've been reading every new take without touching the tools, this is a useful nudge to change the ratio.
He also said that learning alongside people putting AI to work beats every other approach, whether you're a recent graduate, a mid-career professional or a CEO. "The more you can take the time to sort of learn to do, be in teams that are doing, the better it is versus, I think, anything else," he said.
The inclusion of CEOs in that list is a telling detail. It suggests the advice isn't aimed only at junior staff. It's a broader claim about how skills get built when a field moves this fast. If you're a mid-career engineer wondering whether you've missed the window, that's the part to hold onto. Nothing in his framing comes with an age limit, though "be in teams that are doing" assumes you can get onto one of those teams in the first place.
The hiring data tells a more uneven story
Here's the thing: Nadella's advice assumes there's a seat to sit in. Software engineering hiring is picking up, but the recovery favors experienced workers. Senior-level roles made up 69.3% of U.S. software development postings in the first quarter of 2026, up from 54.5% in early 2019, according to Indeed data. That's a sharp tilt toward experience, and it narrows the early-career path Nadella is describing.
Let's be precise about what that figure says. A rising share of senior postings can happen when junior postings hold steady, when they fall, or when total postings grow or shrink. The share alone doesn't tell you which of those is happening. That's the detail you'd want before drawing firm conclusions, and it's the detail that tends to get lost when a single percentage travels through headlines.
Experienced hires also tend to be the people who can catch what automated tools get wrong. Mentoring newer engineers is slower and costs more in the short run, and when hiring tightens, that work often slips first (and this is the part most people overlook). Less training time, less risk, same output. On paper, anyway. The cost shows up later, when the pipeline of people who could eventually fill those senior roles thins out.
That said, Nadella put part of the responsibility on employers. Companies, he suggested, should give junior engineers more chances to build skills as AI reshapes the work. "I care a lot about, hey, can you have an early-career person come in, join one of these teams, go up the learning curve," he said. "So what's the new form of that is what I think we have to really discover."
Notice that he didn't claim to have the answer. That's either honest or a convenient way to avoid specifics. I lean toward honest, but the lack of a concrete plan is the weak point of his argument. Telling employers to rethink onboarding is easy. Figuring out what onboarding looks like when the core tasks are changing under everyone's feet is the hard part.
Other leaders are pushing back on the entry-level squeeze
Not everyone sees the junior talent problem the same way. Amazon Web Services CEO Matt Garman said on The Big Interview podcast last year that cutting off entry-level hiring would undermine the talent pipeline companies need. "If you have no talent pipeline that you're building and no junior people that you're mentoring and bringing up through the company, we often find that that's where we get some of the best ideas," he said.
Garman also challenged the assumption that experience automatically means better AI use. "Many of the most junior folks are actually the most experienced with the AI tools," he said. "So they're actually most able to get the most out of them."
That's a more interesting claim than it first appears. If it's true, the usual logic flips. Juniors aren't just the people who need the tools explained to them. They may be the ones who already use them most naturally. Whether that holds across companies is an open question, and Garman's comment doesn't include data. Still, it's a useful counterweight to the idea that seniority and AI fluency always travel together.
IBM CEO Arvind Krishna has also rejected the idea that AI means fewer openings for graduates. He told CNN in October 2025: "I expect we are probably going to hire more people out of college over the next 12 months than we have in the past few years."
Read that quote carefully. A CEO forecasting his own company's graduate hiring is a statement about one firm, not a read on the labor market. That's not a knock on Krishna. It just means three high-profile executives pushing back on the entry-level squeeze don't add up to evidence that the squeeze is over. They're a counterpoint, and a counterpoint is worth something only when it's weighed against the hiring numbers.
What this means if you're working in this space
If you're an engineer trying to get traction, the practical takeaway is narrower than the headlines. Nadella's advice is about getting close to work that uses AI, not about any specific tool or credential. Look for teams that are shipping, even if the role is adjacent to your title. Build something small and real, and be honest about what it does and doesn't do.
If you're an employer or a manager, the signal is different. Garman's point about pipelines and Krishna's hiring forecast both suggest that cutting junior roles is a bet on the future, and it's one that's easy to make quietly and hard to reverse. You'll feel the effects in a few years, not this quarter.
My read is that Nadella's advice holds up for individuals, but it works best when employers keep the entry-level door open. If that door keeps narrowing, "be in teams that are doing" gets harder to act on. The signal to watch is whether junior hiring rebounds alongside senior demand. That will show whether learning by doing is still a realistic path into the profession, and it's the number I'd check before deciding how much weight to give any single executive's view.