Programming Windows: One Billion (Premium)

Confusion reigned in the wake of Microsoft’s consumer-focused Windows 10 event in January 2015. There was a lot of good news, not the least of which was that Windows 10 would be a free upgrade for Windows 7 and 8.x users during the first year of availability. And there was some bad news: Windows RT was being put out to pasture and would not be upgraded to Windows 10, leaving the future of Windows on ARM in doubt. But there were a lot of unanswered questions too.

Terry Myerson had seemed to indicate that users would be forced to let Microsoft keep their PCs continually upgraded as a condition of getting Windows 10 for free. But even more confusing was the way he framed the support lifecycle. “Once a device is upgraded to Windows 10, we will be keeping it current for the supported lifetime of the device,” he had said.

That was not how Windows support had ever worked. To date, each major Windows version was supported for ten years, a timeframe split into a five-year mainstream support period, during which Microsoft could provide functional updates, and a five-year extended support period, during which it would only supply bug and security fixes. Because the firm typically released a new version of Windows every three years, the support lifecycles for the previous two or three versions would overlap, giving customers the option to stay with whatever version they were using or, if they wished, they could pay to upgrade to a newer version.

Windows 10 seemed to change this. There was no promise of ten years of support—though several years later, Microsoft would quietly revert to this system—and with forced upgrades and two new versions of Windows 10 appearing every year, it appeared that customers—especially business customers—would be spending a lot of time upgrading and rebooting their PCs. But that line about the “supported lifetime of the device” was particularly troubling. It didn’t seem to mean anything.

Sidenote: Within two years, Microsoft would use this line to justify ending support for Windows 10 on PCs that ran specific Intel chipset versions. This was actually done to punish Intel, as the two companies were at the time in a dispute over which company should upgrade the firmware on Microsoft’s Surface PCs. But even this incident did little to clarify the support policy for Windows 10.

Instead of explaining itself, Microsoft simply pushed forward: it released Windows 10 Technical Preview 2 in the wake of the consumer event, publicizing the many functional changes it had made in response to feedback from the Windows Insider Program. And this build included some of the new features Microsoft had shown off, like Cortana integration, the new Xbox app, and the full-screen Start experience for tablets.

There were also hints of trouble to come. With Windows 10, Microsoft would be much more aggressive about pushing its own apps and services, and this version of the OS would come with multiple entry points for “suggestions,” with pop-up notification toasts, Start menu entries, and even File Explorer-based advertising. And the Windows 10 taskbar was populated with some superfluous new user interface elements, like a large search bar and Cortana and Task view buttons. At least they could be removed.

In March, Terry Myerson announced that Microsoft would ship Windows 10 that “summer,” which was a pretty aggressive timeframe given that the first public version had appeared less than six months earlier. The firm even requested that its business customers consider just allowing the in-place upgrade rather than reimaging them as would usually be the case. This suggested that the software giant was confident in the upgrade process, but also that it didn’t understand how its business customers did things.

Microsoft also announced the Windows 10 hardware requirements in March, and things weren’t changing much since Windows 8.1: Windows 10 would require UEFI firmware with Secure Boot enabled, 1 GB or 2 GB of RAM depending on the configuration, and 16 GB or 20 GB of storage depending on the configuration. A Trusted Platform Module (TPM) 1.2 or 2.0 chip was optional, though Microsoft noted that TPM 2.0 would be required one year later.

The firm also revealed that it was once again renaming its modern app platform, which had most recently gone by the name “universal apps.” (Previous names included Metro, Modern, Windows Store, and Store apps.) Going forward they would simply be called Windows apps.

“In Windows 10, we have this notion of a universal app platform,” Microsoft distinguished engineer Don Box said at WinHEC 2015 that month. “And the apps that target it are called Windows apps. Sometimes we say universal apps, but we call them Windows apps. A Windows app can run on every device family: phone, PCs, Xbox, IoT, and other devices like HoloLens.”

“On PCs, we continue to support the two decades-plus worth of Windows desktop applications … for running them on PCs,” he continued. “So sometimes we will talk about a Windows app [what we have been calling universal apps to date] and Windows desktop apps. Windows apps run on all devices. Windows desktop apps [are] PC only.”

Sidenote: curiously, this vague name stuck, though the term Windows apps now pretty much applies to any kind of app that can run on Windows. For example, the Windows App SDK, which would replace UWP years later, is designed to create desktop apps, not mobile apps.

In April, AMD president and CEO Lisa Su said that Windows 10 would launch in July, which was of course within Microsoft’s summer timeframe. She would later be proven correct. But in the short term, our attention was dialed into Build 2015, the conference where Microsoft would expand on its plans for developers who wished to create Windows 10 apps.

And Microsoft kicked off the show going with an explosive and controversial prediction.

“We are going on record, this is the goal,” Microsoft executive vice president of operating systems Terry Myerson told me before the show. “Our ambition is to see one billion Windows 10 devices within two to three years. No other ecosystem has this reach.”

One billion Windows 10 devices. Within two to three years.

It was an audacious goal. It was also next to impossible.

Microsoft had recently touted the size of the Windows PC installed base as 1.5 billion units, and Microsoft/Nokia had sold over 100 million Lumias since the launch of Windows Phone 7 in 2010. (Lumia accounted for the majority of Windows Phone sales, but this figure was likely double that of the actual installed base at the time.) Other Windows 10-based platforms, like Xbox One, sold in negligible quantities, while others, like Surface Hub and HoloLens, had yet to ship but were expected to be niche products with low unit sales.

So, how would Microsoft get to one billion?

The easiest way was to simply ship Windows 10 on new PCs: Microsoft’s hardware partners had sold just under one billion PCs in the previous three consecutive years, but that wasn’t a good measure of how many copies of the latest version of Windows were installed on those PCs. For example, Microsoft had sold 20 million Windows 7 licenses per month consistently in the three-year period before that, for a total of about 720 million licenses sold in three years; in that three-year period, hardware makers had sold over 1 billion units. If Microsoft and its hardware partners could somehow repeat the success of Windows 7 with Windows 10, that was a sold way to get over 700 million Windows 10 devices into the world within three years.

Sidenote: That wouldn’t happen, as the PC industry hit a modern low point for sales between 2015 and 2020, and just over 800 million PCs were sold in the three years after Windows 10’s launch.

700 million PCs sold plus about 50 million Windows Phones, 10-20 million Xbox Ones and some single digit million units of Surface Hub and HoloLens combined, and Microsoft was still looking at a deficit of over 200 million units. And that figure could be made up by free Windows 10 upgrades. That is the real reason the Windows 10 upgrade was free: to drive Windows 10 usage and help Myerson hit his goal of one billion Windows 10 devices.

Sidenote: What Microsoft wasn’t counting on was new Windows Phone sales. But at this point, the firm had already started the process of dismantling that business, and while it wouldn’t communicate this publicly for years, it knew on the day of the one billion announcement that Windows Phone would not play a major role in achieving this goal. Myerson confirmed this to me when he left Microsoft a few years later.

The Build 2015 keynote opened with a preview of how CEO Satya Nadella would reorganize Microsoft’s core business units into the intelligent cloud, the reinvention of productivity and business processes, and more personal computing. That the segment about intelligent cloud—mostly Azure—came first and went on for a monotonously long time was likewise a preview of how the firm would prioritize Windows going forward. Despite the new Windows 10 platform, it would be given third billing, and Build 2015 would be the first Microsoft developer show in decades where Windows wasn’t the primary concern.

As such, attendees were forced to wait an interminable hour and thirty minutes before Terry Myerson finally appeared to discuss Microsoft’s plans for Windows 10.

Expanding on Nadella’s January comments about Microsoft wanting customers to love Windows 10, Myerson claimed that the software giant also wanted developers to love Windows 10. And it would do so by making Windows 10 its most attractive developer platform ever, thanks to its broad support for a diverse range of form factors.

“With Apple, you choose to invest in iOS or macOS,” he said, drawing a comparison with the platform maker than had just surpassed Windows by userbase. “With Google, you choose to invest in Android or Chrome OS. Windows is the only ecosystem that lets you [write a single app that works everywhere].”

As with the mobile platforms with which it was trying to compete, Microsoft offered a mobile app store called the Windows Store. This store came with “promises to the user,” Myerson said, related to security and reliability, and though this wasn’t a true differentiator with the Apple App Store or Google Play Store, this was still important because most Windows users still used the web to find and download apps. And there would be a version for businesses as well, he noted, offering IT a way to curate the apps its users saw.

Terry then previewed four new ways that developers would have to bring their apps to “one billion Windows 10 customers,” a curious description given that Windows 10 wasn’t even available yet, let alone used by such a large audience. It would let developers bring “server-hosted websites,” desktop applications with .NET and Win32 codebases, and Android phone apps built with Java and C++ to the Windows Store as apps. (The Android app support would require Microsoft to build an Android subsystem for Windows, Myerson said, though this would never appear.) And iOS developers could bring Objective C codebases, but not entire apps, to Windows as well, though that porting process would be more involved.

Sidenote: there were over 16 million Windows desktop applications in existence at the time of this talk.

Myerson then introduced Joe Belfiore, who launched into yet another Windows 10 demo, this time focusing on platform updates that would make it easier for users to engage with apps. The visuals and overall design had been improved since the last public demo, of course, and Microsoft had brought back some features users had missed from Windows 7, like jump lists. Microsoft would suggest apps in the Start menu, he said, and when users did install an app from the Windows Store, that app would be promoted in the Start menu so it could be easily found.

Belfiore showed off Windows Spotlight, a new lock screen feature that would provide high-quality new wallpaper each day, new Cortana features, the new Microsoft Edge web browser, and Continuum, which was coming to Windows Phones as well. Then, Alex Kipman appeared to demonstrate HoloLens, Microsoft’s Windows 10-based augmented reality solution.

On day two, we finally learned more details about the Windows 10 developer story. As was the case with Build 2014, this year it fell to corporate vice president David Treadwell to communicate the way forward. And as was the case with Build 2015, it would center on universal apps, which would now run on Surface Hub and HoloLens in addition to PCs, tablets, phones, and Internet of Things (IoT) devices. And once again, he started off with an insider story to pique the audience’s interest.

“I’d like to tell you about a conversation that Terry [Myerson], Kevin Gallo, and I had last summer with one well-known developer, a guy named Bill Gates,” he started. “We were reviewing with Bill our plans to make Windows 10 great for developers like you. As you might imagine, a review with Bill is always a big deal, so we were well prepared. But we were also nervous. One of the points that I made is that we’ve been getting lots of feedback from people like yourselves, that they didn’t like the Modern platform in Windows 8. And that’s that it really only supported full-screen applications. No windowing.”

“Full screen is great on a smaller screen, like a tablet,” he continued. “But many people like large screens [or] multi-monitor setups. I love the three-monitor setup that I have. So I told Bill, one of the things we’re doing in Windows 10 is we’re bringing back this find feature from 1986: resizable, overlapping windows for apps.”

Here, the audience broke out into applause though, by this point, Microsoft’s plan for windowed Store apps was well-known.

“And Bill paused,” he said, “and he looked at me kind of intently for a second, and then he said, [sarcastically] ‘Wowww! That’s a really great idea. You know what we should do? We should name the product after that feature’. [Laughter.] So I told him that the joke in the engineering team is that we really should have called it Window 8.”

This time, the laughter broke into applause, despite the obviousness of the joke. Treadwell had accomplished something important: he had connected with the developer audience using an internal story about Microsoft’s most revered figure, and he had not just distanced Windows 10 from its predecessor, but had in fact mocked that predecessor and, by implication, the people who had created it. People who were now all gone from the organization building Windows 10.

“That story does show something important,” he concluded. “While it is valuable to have experiences that span multiple devices, we as developers must be the user at the center of these experiences. What works great on a tablet may not work great on a huge, modern desktop monitor. This important idea from our conversation with Bill carries through both of the developer conversations that Kevin and I are excited to talk about with you today.”

The first of those two conversations would include how Windows 10 would provide a “continuum of experiences” across different device types that put the user at the center. The second was how Windows 10 would let developers take existing codebases and reach over one billion Windows users via what was now called the Universal Windows Platform, or UWP.

Sidenote. Remember that the then-current installed base of PCs that could run UWP apps was, in fact, zero at that time: this system would require Windows 10, and would not be backported to Windows 8.x. But even if one included the Windows 8.x user base, that only included about 200 million PCs.

As an evolution of the previous universal Windows apps platform, UWP would let developers target a broader range of device types than ever before, Treadwell noted, including, he said, Raspberry Pi, a platform that wasn’t included on the slide. (Technically, the Raspberry Pi implementation of Windows 10 fell under the IoT umbrella, as it didn’t include an end user interface with a desktop and app windows.)

Treadwell then launched into an update to his 2014 talk, describing how Microsoft had made improvements to each phase of the Windows app development process—design, develop, debug, and distribute—with UWP. It would provide a single design language, he said, and one set of controls across the platform. This was, in its own way, also a refutation of the Sinofsky years: where Sinofsky had stolen some ideas from Windows Phone and reimplemented them differently in Windows 8, the same set of controls, including those that had originated in Windows Phone, would now appear on every platform supported by UWP, including Windows 10 for PCs.

The ability to reuse existing code so broadly was a major step forward. But targeting so many platforms with a single codebase had been a holy grail of sorts for the industry for decades, an elusive dream. So at a high level, UWP was very interesting indeed, though critics would quickly grumble that a truly universal platform would target non-Microsoft platforms as well.

Sidenote: Microsoft would take steps to address that need a year later by acquiring Xamarin, which produced tools and .NET-based technologies that let developers write a single codebase that could target Windows (UWP), Android, and iPhone. But this product, which was called Xamarin.Forms and would later evolve into .NET MAUI, would always remain separate from UWP.

One of the big conceptual hurdles with UWP was how apps written to so many different platforms could possibly be tailored, as Microsoft called it, for each platform. To this end, the firm had created so-called action triggers that would automatically adjust an app’s layout using familiar XAML markup. Tailoring app experiences wasn’t just about UI, however: you could also tailor based on input, orientation, sensor availability, services, or many other detectable changes. If all else failed, you could even create a custom XAML view for each device type, each of which shared common code.

With UWP, Microsoft had also expanded on the ability to create a single app that worked across multiple form factors by adding a packaging format—called AppX; it had actually debuted in Windows Phone 8.1—that would likewise work across all supported platforms. That streamlined the build process, since Visual Studio wouldn’t have to package apps differently for different platforms. And it dramatically simplified distribution, since that single install package would work everywhere that UWP worked.

“The Universal Windows Platform is really just [version] 10 of the same modern platform you’ve been using on phones since Windows Phone 8.1 and on PCs since Windows 8,” he continued, inverting the actual timeline. “Your code still works [and] your concepts still hold true. You can continue writing apps to the Universal Windows Platform in your language of choice, whether that’s C, C++, C#, Visual Basic, or JavaScript. The Universal Windows Platform brings together the best of Windows, phone, and Xbox.”

Except, of course, that it did not.

The problems with UWP were many, but the central issue was that UWP was a mobile app platform, and not a desktop app platform. And that meant that while developers could write apps that would work across multiple platforms, it also meant that those apps would be mobile in nature and would thus lack the sophistication and power that those 1.5 billion Windows PC users expected.

Worse, the hard work that Microsoft had done to make UWP “universal” within its own platforms would never pay off: that year, Microsoft began the slow and painful process of dismantling Windows Phone as a platform, taking away a key benefit of UWP to developers. And the remaining non-PC platforms—IoT, Xbox, Surface Hub, and HoloLens—would always be niche markets, at least for UWP developers, and few would ever see the point of porting their apps to these less desirable form factors.

Microsoft’s push to continue forward with UWP was in some ways understandable: of the many things it had done right in creating and maintaining Windows as a platform, it had almost always provided developers with a way forward as it evolved. But the big mistake had been made with Windows 8: instead of basing its new developer platform on the powerful and capable .NET frameworks of the past, that team was overly afraid of the mobile wave that was then taking over personal computing, and it made the wrong decision to bolt a mobile app framework—one not based on .NET—on top of Windows 8. UWP would simply carry this mistake forward.

One might wonder why Microsoft continued forward with the same mistake. But when UWP was conceived, the Windows Phone team had just taken over the Windows group, and they were quite interested in getting feature and API parity with “big” Windows where that made sense. And this group at first saw Windows Phone, and not Windows, as the volume platform of the future, and UWP was positioned to accommodate a coming time when mobile apps, not desktop apps, were more important. As noted above, of course, that never happened.

But even if you just consider UWP as a platform for writing Windows 10 apps, there were other problems, too, problems no one saw on that hopeful day in April 2015. Each version of UWP would be tied to a specific version of Windows 10, erasing the defragmentation gains that Terry Myerson had promised. What this meant was that new UWP features would only work on a specific version of Windows 10 or newer; previous versions, though still supported, could not run apps that used those new features. And with new versions of Windows 10 appearing twice each year, the fragmentation only worsened as time moved forward, until there were more supported versions of Windows than had ever been the case in the past. It was a problem a post-Myerson Windows team would finally resolve, but not until several years later.

At the conclusion of Treadwell’s presentation, Kevin Gallo provided a demonstration of what it would be like to create a cross-platform universal app using Visual Studio 2015 or, as he called it in deference to Microsoft developer events past, a “lap around the Universal Windows Platform.” Creating a single app that made sense across such a diverse range of form factors—a “plethora” of devices, Gallo called it—was, of course, difficult. But Microsoft’s history is rich with such demos, and what he really showed was how you might use Microsoft devices in the design phase of new app development. When he finally got to the source code, the app had, of course, been created in advance and so Gallo highlighted code segments as he spoke, calling out new features in UWP.

Treadwell returned to discuss the debugging and app distribution advances that Microsoft would add to Visual Studio 2015 and Windows 10. And then he turned to the so-called “bridges to the Universal Windows Platform,” Microsoft’s plan to get developers to port code that they had written for apps on other platforms—Android, iPhone, whatever—to Windows 10.

“Start with your code, no matter where it was born, and bring it to the Universal Windows Platform and Store,” he said, as a slide noted there would be bridges for the web, .NET and Win32, Android Java/C++, and iOS/Objective C.

Each bridge would offer a different level of support. For example, the new EdgeHTML engine in Microsoft Edge would let developers reuse their existing website code by creating and packaging an app shell of sorts that simply pointed to the web URL of that site. For those developers who had stuck with “classic desktop technologies” like Win32 or .NET, Microsoft would let them package their applications with AppX. Android and iOS developers would likewise be able to import their code, though the Android “tooling,” as Treadwell called it, was suspiciously not ready yet. But in each case, these ported apps could be distributed from the Windows Store, and they could be extended with native Windows 10 features like Cortana, Xbox Live services, inking, notifications, and much more.

In the wake of Build, a one-liner from one Microsoft engineer in one Build session started making the rounds until it became canon: Windows 10, he said, would be the last version of Windows.

“Because Windows 10 is the last version of Windows, we’re all still working on Windows 10,” Jerry Nixon said during a talk about tiles, notifications, and the new Windows Action Center. This comment was never repeated by anyone else at Microsoft, let alone an executive or decision maker. But it set off a firestorm of controversy that continued through the time Microsoft announced Windows 10’s successor, Windows 11, in 2021. And Microsoft was forced to issue a statement about the line, noting only that its Windows as a service vision meant that it was planning to keep upgrading Windows 10 going forward with functional updates, and security and bug fixes.

“We aren’t speaking to future branding at this time, but customers can be confident Windows 10 will remain up-to-date and power a variety of devices from PCs to phones to Surface Hub to HoloLens and Xbox,” the statement explained. “We look forward to a long future of Windows innovations. Windows will be delivered as a service bringing new innovations and updates in an ongoing manner, with continuous value for our consumer and business customers.”

In May 2015, Microsoft announced that it would sell Windows 10 for PCs in four core editions, Windows 10 Home, Windows 10 Pro, Windows 10 Enterprise, and Windows 10 Education, but it declined to reveal pricing because, well, almost anyone could get it for free.

Then, on June 1, we learned that Microsoft would ship Windows 10 for PCs on July 29.

“We designed Windows 10 to create a new generation of Windows for the 1.5 billion people using Windows today in 190 countries around the world,” Mr. Myerson noted in a blog post announcing the date. “With Windows 10, we start delivering on our vision of more personal computing, defined by trust in how we protect and respect your personal information, mobility of the experience across your devices, and natural interactions with your Windows devices, including speech, touch, ink, and holograms.”

Microsoft then started promoting the coming free upgrade to Windows 7 and 8.1 users via a little Windows flag icon that would appear in users’ system trays. Clicking it would launch the “Get Windows 10” reservation app, which would let one “reserve” the free upgrade and, more to the point, provide Microsoft with its first real data about the public’s interest.

Windows 10 would be a major upgrade over Windows 8.x, but it would also mark the end of the road for some aging—and, in some cases, beloved—technologies: DVD playback, Windows Media Center, desktop gadgets, floppy drive support, and other Windows 7/8.x features would be deprecated or even removed when customers upgraded to Windows 10.

With Windows 10 set to be finalized any day, things started changing rapidly. Microsoft dropped a requirement that users initially configure Windows 10 with a Microsoft account (a requirement that would reappear in Windows 11 years later). It announced major changes to the Windows Insider Program to set up the program for the post-RTM world. New builds focused more on quality and less on new features, and the final “hero” image—which Microsoft would use in marketing and as the system’s default wallpaper, was revealed.

And then, on July 13, Terry Myerson appeared at the company’s Worldwide Partner Conference in Orlando to announce that Microsoft had finalized Windows 10 for PCs. Except that it hadn’t because of some last-minute quality issues. And so Terry simply spoke about some of the advances in Windows Mobile 10 and Windows 10. Two days later, however, Microsoft shipped Windows 10 build 10240 to Insiders, which my sources had told me was the final build.

“This build is one step closer to what customers will start to receive on July 29,” Microsoft’s Gabe Aul wrote in the announcement post. “Builds from here on will only be available through Windows Update … Remember that you’ll need to be logged in with your Microsoft Account in order to receive the build. Besides builds, over the next two weeks you’ll also see some Windows Updates and app updates in the Store, so make sure to keep checking for updates daily to make sure you’re running the latest and greatest code.”

That week, Amazon posted preorders for Windows 10 Home ($120) and Windows 10 Pro ($200) physical media, which consisted of a USB drive, not an optical disc, and a reported release date of April 16. But Microsoft, curiously, refused to acknowledge that it had completed Windows 10. Instead, it simply shipped build 10240 to PC makers with the understanding that some updates would be made available between then and July 29, and then again after that, as Windows 10 was a moving target that would always be kept up-to-date. (The first major Windows 10 update, called Service Release 1 (SR1) internally, was released just a week after Windows 10 arrived.)

Finally, July 29 arrived. Microsoft issued the final ISO files for Windows 10, proving that 10240 was the final build. The first Windows 10 reservations started downloading to customers’ PCs, letting them upgrade from Windows 7 or 8.1 for free. And I published my review.

“Windows 10 is ideally suited for every PC form factor imaginable: traditional PCs, yes, but also touch-first devices and, most decisively, the 2-in-1 PCs that can move between these usage modes,” I wrote. “Windows 10 will make any PC—desktops, laptops, 2-in-1s, tablets, and mini-tablets, whatever—better, and that’s not something that could be said of Windows 8. I honestly didn’t think it was even possible. Start is again a menu on PCs, as it should be. Non-discoverable UIs are gone, banished to the same tech dustbin where you’ll find Microsoft Bob, Clippy, Kin, and other Microsoft experiments gone wrong. [And] where Windows 8 alienated users, Windows 10 embraces them, with Microsoft apologizing and saying we’re listening again. It feels good, sure. But it also works, which is more important. Like Windows phone before it, Windows 10 isn’t just different to be different, it’s just better. And it’s not mindlessly fulfilling all users’ wish lists. It does the right thing.”

“Windows 10 is superior to both Windows 7 and Windows 8.1, and its user experience works as well or better on the different Windows device types—traditional PC form factors for Windows 7 and ‘touch-first’ devices like tablets and 2-in-1s for Windows 8.1—than do its predecessors,” I continued. “But Windows 10 also exceeds Windows 7 and 8.1 in other ways, with an evolved universal apps platform that all users (and developers) can embrace, cross-platform chops that are unparalleled on other mobile and desktop computing systems, and an adaptable, user-focused user interface that can be customized to your liking at every step of the way. And it respects the way you work, whether that’s keyboard and mouse, touch, pen/stylus, or any combination of those things. I am ecstatic to see the Windows team finally listening to its customers, and it has delivered a wonderful gift here, especially to those who felt that Windows 8 did not meet their needs. In combining the best of Windows 7 with the best of Windows 8.1, Microsoft has turned the page on a disappointment and done right by all of us.”

In a little over 30 days, Microsoft would announce that it had upgraded over 75 million PCs to Windows 10 across 90,000 unique PC or tablet models. While not directly comparable, that figure exceeded the 60 million Windows 7 licenses that Microsoft said it had sold in the first 60 days of that system’s availability in 2009. And it almost doubled the figure for Windows 8, where Microsoft said it had sold 40 million licenses in the first month.

A new era had begun.