Microsoft Gained $25 Billion in Market Value Today. Why Is a Company Drifting Further From Us Doing Better and Better Business?

Two and a half years into Satya Nadella's tenure, Microsoft's market cap is up 50%

Microsoft reported its earnings for the third quarter of the year this morning. Cloud computing and Office drove the results; the stock rose 5.9% in after-hours trading and the company's market value gained more than $25 billion — roughly what it paid for LinkedIn, earned back in an evening.

On the earnings call, one analyst asked: "Microsoft wants to do artificial intelligence — how? How will it tie into the business going forward? Where is the investment going?"

He was asking about Microsoft's latest reorganization. In late September, Nadella launched a new round of restructuring. Qi Lu, who ran the Office and Bing businesses, left the company. Microsoft announced a new Artificial Intelligence and Research group, with more than 5,000 people working on AI products and the underlying technology.

Unlike tech-company leaders such as Jeff Bezos or Elon Musk, Nadella said nothing memorable. He simply ran through what Microsoft plans to do with HoloLens, Skype, and processors, and said the company would turn long-term research into finished products.

The public speeches and company-wide memos of this Indian-born CEO have always been this flat — boring, to put it plainly. Nadella would never do what his predecessor Steve Ballmer did: yell "Developers, developers, developers!" at thousands of people over and over, until his shirt was soaked through with sweat.

But he has clearly gotten a lot of things right. Look at the stock market.

In February 2014, when Nadella took over, Microsoft was worth less than $300 billion. Today it is worth about $445 billion — nearly half again as much.

Over those two and a half years, Nadella has cut more than 20,000 jobs across several rounds of layoffs. Look at where the revenue now comes from and Microsoft is a different company.

Productivity and Business Processes is the Office business. Besides the traditional Office software, it includes the 24 million individual consumers who pay Microsoft every month to use Office 365. Today's Office runs on the platforms of rivals: iPad, iPhone, Android.

And the cloud business supports every flavor of Linux.

Windows has little to do with either of these two segments, which together supply more than 70% of Microsoft's profit.

When Ballmer Left, Microsoft Looked Nothing Like This

In August 2013, when Ballmer abruptly announced his retirement, Microsoft was at a low point. Apple and Google were strangling its smartphone business. Cloud computing was Amazon's market. Bing search kept burning money. Windows 8 was the most criticized operating system in the company's history. And Office was still chained to Windows.

As it had for two decades, Microsoft had a foot in every corner of the computing market, and was doing well in none of them.

Two big projects Ballmer had set in motion were still under way.

Talks to buy Nokia's handset business had reached their final stage: Microsoft was about to start making its own hardware.

Microsoft had also announced a reorganization called "One Microsoft."

Until then, Microsoft was split into five businesses, each of which disclosed its own sales and profit every quarter. They were, in effect, five small companies — each answerable for its own targets, each buying its own advertising, each with its own marketing department to sell its own products.

Under that structure, whoever made the most money had the loudest voice. After the iPhone and Android arrived, for instance, Microsoft produced a modern mobile operating system, WP7, in 2010. It then went nowhere for two years, because the Windows division had its eye on the phone business and insisted on freezing development of the phone system so it could be updated alongside Windows 8.

Those two years of waiting buried the last hope for Microsoft's phone platform.

Ballmer hoped a reorganization would stop that kind of conflict from happening again: five engineering groups would handle development for Office, Windows, cloud computing, internet services, and hardware, while marketing, business development, and finance would be run by unified teams — and the company would no longer report each business's profit or loss directly.

The plan drew doubts from the start, because Microsoft's businesses were too different from one another.

• Office, sold mainly to companies, for employees' daily work;
• Windows, licensed to PC makers to install on their machines;
• Cloud computing, sold to companies to run their operations;
• Hardware — Xbox, Surface, Windows Phone handsets — sold straight to consumers;
• Internet services, essentially a smaller Google, living off advertising.

Every division's products ran on a different business model. Some needed to persuade business owners; others had to win consumers one at a time. A single marketing team could hardly serve needs that different. Microsoft did not have a unified brand image then, either.

A month into his reorganization, Ballmer announced his retirement. Only then did Microsoft begin looking for CEO candidates.

His departure was clearly not entirely his own choice. Information that emerged later showed that Bill Gates, Microsoft's principal founder, had opposed both the reorganization and the Nokia deal.

The Reorganization Ballmer Started Was Finished by Nadella — a Different Way

On February 4, 2014, Satya Nadella, who ran the Cloud and Enterprise division, was named CEO by Microsoft's board. Founder Bill Gates also returned to the company as a technology adviser.

Nadella joined Microsoft in 1992. For more than a decade he worked in online services and in the server division — the latter turned, in 2010, into what is now Microsoft's Cloud and Enterprise division.

Picking an obscure executive from the inside was, without question, the conservative choice. Until then he had looked less like a man with an agenda than like a man who executed the goals Gates and Ballmer set.

Before he became CEO, Nadella told a reporter: "If they" — Gates or Ballmer — "are screaming at you, telling you you're crazy, that you're going to destroy the company, you have to keep proving your idea with data, over and over. Doing that also clears up your own thinking."

Even Microsoft's own customers saw it that way. Tim Theriault, chief information officer of Walgreens, the largest drugstore chain in the United States, said Microsoft had trapped itself.

Nadella did not "disappoint," either. The open letters he wrote to the whole company after taking office were stuffed with vague corporate slogans, and he promised to keep investing in mobile even though Windows Phone plainly had no hope left.

But what he said mattered far less than what he did.

In July 2015 came the second round of layoffs from before Nadella took office: 7,800 jobs cut, and with them essentially all of the Nokia business Microsoft had bought.

After that, Nadella kept pushing phones — and Windows itself — toward the margins.

A month before those layoffs, in the name of efficiency, he folded the hardware business into the Windows division and pushed out every former Nokia executive, Stephen Elop included. The Windows division suddenly carried hardware's losses, and this year it finally killed the entire phone hardware business.

Internet services was the other business Nadella weakened.

To fight Google for search and advertising, Ballmer hired Qi Lu away from Yahoo in 2008 to run online services. After that, all of Microsoft's internet services business revolved around Bing, and burned a great deal of money.

In June 2015, while Nadella was cutting into hardware and Windows, the internet division sold its 100-person Bing Maps team to Uber. Its main work became buying up apps and games and rebuilding them as Microsoft tools — almost none of which supported Windows.

• December 2014: bought the email app Acompli, later turned into mobile Outlook;
• February 2015: bought the calendar app Sunrise, later folded into mobile Outlook;
• May 2015: bought the to-do app Wunderlist;
• November 2015: bought the mobile data-analytics app MileIQ;
• March 2016: bought the phone keyboard SwiftKey;

Every mobile app Microsoft bought over the past year or two was built for two platforms: iOS and Android.

Internet services, once built around Bing, was rebuilt into a division that supplies productivity tools on iPhone and Android phones.

In late September 2015, Microsoft changed how it reports its earnings. From then on its main businesses were Office, cloud computing, productivity apps, and the Surface line of computers — every one of them aimed at making work more efficient.

This is Microsoft's new image.

Early last year, for the first time since Windows 95, Microsoft ran an ad in the Super Bowl, America's answer to the Spring Festival Gala. In line with the Windows 10 tagline, "a more human way to do," it told a story about technology changing lives.

Off the field, viewers also saw the Surface Pro 3 used for tactical analysis during the broadcast.

Ballmer's "One Microsoft" reorganization was finished by Nadella in the end — by a route nothing like the one Ballmer had imagined.

The Question Is Whether AI Can Bring Microsoft Back Into Consumers' View

The rebuilt Microsoft has one image and is a more efficient money-making machine. It is also more boring. Windows Phone was abandoned outright, Windows is fading, and Microsoft has become, entirely, a company that sells to companies.

For the past two years, the only bright spots have come out of Microsoft Research, the part of the company that does long-term work and publishes papers.

At the Windows 10 launch, the most exciting moment came when Alex Kipman, the inventor of Kinect, demonstrated the augmented-reality headset HoloLens.

In the middle of last year, users on social networks around the world were showing off their guessed ages. A site called how-old.net took off without collapsing under the traffic. It turned out to be face-recognition technology built by two Microsoft engineers on the Azure cloud platform. Behind it sat a machine-learning effort called Project Oxford which, besides image recognition, was also working on speech recognition.

And the chat AI that had been live in China for more than two years had 3.97 million followers on Weibo alone.

In March this year a similar product, Tay, arrived on Twitter and Facebook in North America, and nearly 200,000 people struck up a conversation with it in a single day. Tay was shut down a day after launch for spreading racist remarks — but Microsoft's earlier and far more expensive internet products could not draw that many people in a single day without advertising behind them.

Now Microsoft has made the artificial intelligence behind these projects its new direction for the future.

In June, in an article he wrote himself for Slate, Nadella compared AI to the internet in 1995.

"I remember reading Bill Gates' memo The Internet Tidal Wave in the spring of 1995. In it he foresaw the technical development and commercialization of the internet in connectivity, hardware, and software. More than 20 years later, we are looking for a new wave — the wave of artificial intelligence…"

On September 28, Microsoft set up its new artificial-intelligence research group, putting a team of 5,000 people to work turning AI into products.

As with every reorganization at Microsoft, this one came with a shift of power at the top.

The day before it was announced, Qi Lu, executive vice president and head of the internet business, left the company, citing his health. He had been closely involved with the Xiaoice project, and had hosted the launch events for two new versions of Xiaoice.

But the man who ended up running the AI group was Harry Shum, the executive vice president who had been in charge of Microsoft Research and had pushed its work from technology toward products.

Today several of America's largest tech companies have their eye on artificial intelligence. Every Google and Facebook product applies it. Amazon has put a team of more than a thousand people on AI software and its smart speaker. Even Apple has begun mentioning "machine learning" and "artificial intelligence" again and again this year, at its events and on its earnings calls.

But if the computing revolutions of the past teach anything, it is that none of these companies may actually get AI right and come to dominate the next market.

In 1995 Gates saw clearly how important the internet was — yet the internet companies that rose were new ones: Google, Amazon, Facebook.

The smartphone market went the same way. Microsoft and Nokia were both taking smartphones seriously in the early 2000s. In the end it was Apple, then worth a tiny fraction of Microsoft, that led the phone revolution with the iPhone.

Even if artificial intelligence really is the next technological wave, can Microsoft win this time?

Images from PCWorld, GeekWire, Microsoft, and TechRadar.