Microsoft helped kick off the AI boom. It needs humans more than ever, its CEO says

The need for more well-rounded employees isn’t just a result of automated software generation. The nature of Microsoft’s next chapter and its broad ambitions to be the foundation for a new era of computing will require more than just code. Enabling autonomous AI “agents” that supercharge human productivity requires the construction of a new system of online interaction that is more complex than the early internet protocols we still use today, but with equally important, long-term implications.

For Nadella, this moment has been an exercise in patience. His first memo as CEO of Microsoft just over a decade ago was about this very moment, which he called “ambient intelligence.” The concept was that the clicking and typing and scrolling that defines technology would eventually give way to computers that operate in the background and carry out mundane tasks automatically. But the technology wasn’t there yet, so Nadella yielded to pragmatism. “Then I said, ah, this is just too crazy and esoteric. Let’s make it simple, and we’ll call it mobile first,” he said.

Despite having his head way into the future, Nadella steered Microsoft to one of the most profitable journeys in business history as it rose in value from $300 billion to $3.4 trillion today, making it the most valuable company in the world.

But roughly five years ago, he gave Scott the green light to build one of the world’s largest supercomputers to train what would eventually become ChatGPT. The move was in service of his plan from day one. “This is the next phase of it. We can lean into it, where computing is everywhere and nowhere, so to speak,” Nadella said.

That phase also means a shift in identity for Microsoft. Today, the first thing every application does when it opens is display the name of the brand of the company that made it. Hardware devices display corporate logos. All of that conveys some kind of meaning to customers. “Microsoft products have this sense of, ″⁣When I use a Microsoft product, I feel good about myself;‘” Nadella says.

But in a world with ambient intelligence, or agentic AI, Microsoft’s goal is to fade into the background while simultaneously becoming a bigger and bigger part of its customers’ lives. It’s a bold idea, but one that requires a kind of humility.

“My way of recruiting anyone who’s coming to Microsoft is saying, ’hey, look, if you want to be cool, go join somebody else. If you want to make others cool, join Microsoft,” Nadella says.

Navigating that shift would simplify the world for Microsoft’s customers, but in some ways make the company’s job much more complex and open-ended.

And it’s a puzzle that requires the input of humans, more than just clean lines of code.

Scott, though, believes being uncool does not mean being uninteresting. And in the AI era, engineers will be defined by their breadth of curiosity and creativity. And to date, that is not the archetypal Silicon Valley coder.

“You can quote me on this: Some of the conversations I have with my people right now are just so goddamn boring,” he said.

And that is the message, in a nutshell, that Scott was planning to convey at Build this week. For some reason, software developers are not taking full advantage of the AI capabilities that exist today — something he calls “capability overhang.”

“There’s going to be a little bit of me pushing on folks at large, the same way that I do on our engineers inside of Microsoft, which is: you should just let your imaginations run wild,” he says.

Scott sets a high bar for what he calls his own pathological curiosity: “I’m curious about everything. I used to get yelled at all the time. ‘Oh, you need to focus more on one thing.’ It’s like, I don’t want to focus on one thing. I want all of it.”

He was standing inside the physical manifestation of his pathological curiosity, a workshop that feels like a combination of Thomas Edison’s lab and a design studio. There are 3D printers, CNC machines, lasers, wire benders and a “soft goods’ studio where he does book binding, and handbags that are sold at a San Francisco boutique.

He also makes Japanese bowls, which have become a metaphor for the role of human labor in the AI era. “I come in every day with a very Japanese mentality,” he says. “I try to get a little bit better at this thing, every day, and I’m doing it because there’s dignity in being better at that thing for myself, independent of everybody else. But also because what I’m doing is in service to other people.”

Nadella is pushing Microsoft along the same path: One in which technical product building is more focused on its end-users than Microsoft, with its enterprise culture, has traditionally been to other people. That could mean the company is viewed as less sexy than its competition, but Nadella sees that as its competitive advantage. “At the end of the day, I think if Microsoft’s going to do something, it’s going to be that platform approach,” he says. “That is the identity of Microsoft. We should be ok with that.”

Through that lens, Nadella sees the shift to AI as simply a new way for humans to do the same kind of knowledge work. It’s Microsoft that has to adapt and compete in order to remain a key part of that work.

He asks me to imagine what an alien intelligence would have thought if it came to Earth and observed white collar workers in the 1980s and then returned today. They would have seen an office with a handful of typists, and others making physical slide presentations while some people send and transport interoffice memos. Upon return, they’d see every single person sitting in front of a computer screen.

“They would say, ‘oh my god the entire human race has become typists!’” he said.

In the AI future, where the agentic web is established and ambient intelligence is the norm, the aliens might be even more confused as humans appear to wander around aimlessly, untethered from physical artifacts. In reality, they’d be witnessing the same thing: Humans going about their work day using the latest tools available.

There are a lot of similarities between what is happening in software engineering and what happened to the media industry over the past two decades.

Publishing before the early 2000s was something that a small group of people had the ability to do. It required infrastructure and training. Every article or book that was published was more or less professionalized, and writers devoted their lives to honing their craft. That all changed when the web and then social media turned everyone into a publisher.

At that moment, the power and profitability of content creation began to diminish. Building the infrastructure, or platforms, that enabled ubiquitous online content creation was where the power and fortune resided.

Until now, building consumer software was similar. Writing code was a skill that took years to hone. Big companies and venture capitalists largely decided which software ideas would get resources and become commercial products. The vast majority of people would never write a line of code in their life and were fine with it.

As we enter a new era where software development becomes ubiquitous, every customer of software companies will quickly become a competitor.

Within that landscape, there is an opportunity to build the infrastructure that enables people to create the customized software that will automate their lives. There may be room for a few of those large platforms.

You could argue that there are more people creating content today than ever before in human history. You could also argue the media industry lost a lot of jobs.

Of course, there are a lot of differences between the two industries. Content creation serves a very different purpose than software development.

Still, it may feel like coding jobs are going away,at least in the short term. When I asked Scott about this — Microsoft isn’t expanding its ranks of software developers right now — he said it’s a transient thing.

And Microsoft trying to become the underlying platform that enables this change is no small undertaking. It’s less a computer science problem than a systems problem. In some ways, it’s like building a new civilization. If software is going to act autonomously and interact on the web with other autonomous pieces of software (what Microsoft is calling the “agentic web”) then we need rules and regulations and norms.

This effort resembles the early development of the web, when all the protocols we use today were conceived. But it’s even more complex and happening much more quickly and globally.

So it makes logical sense that it is going to require a lot of human effort to build it and maintain it. If you want to call those people software developers, fine. Most importantly, they’ll be people.