Then and Now, Part 2: Feeling Blue (Premium)

Steven Sinofsky was right: It wasn’t enough for Microsoft to just port Windows to Arm, the mobile chip architecture behind Android and Apple’s iPhone and Android. For this new platform to succeed, Microsoft would have to aggressively remove legacy code, develop a modern apps platform, invent versatile new form factors, and attract developers and customers to something new that offered a bridge to the PC past. And it would have to happen in just three years.

In the end, Sinofsky and the software and hardware teams he oversaw were thwarted by a combination of factors. The most problematic, perhaps, came from within. Windows RT was undermined from its inception when the Microsoft Office team, then distracted by making versions of its apps for Apple’s iPad, refused to fully support the new app platform, called Metro. It was immediately clear that this decision would send a message to Microsoft’s customers and, as important, to developers: If Microsoft didn’t support this new platform fully, why should they?

And so he did the only thing he could do: He compromised. Instead of being an entirely new system that could take advantage of the rich capabilities of its Windows foundation but on more reliable and efficient devices, RT was instead a confusing version of the traditional Windows 8 product for PCs that only ran on Arm devices and couldn’t even run Windows desktop apps. It looked identical to Windows 8. But it wasn’t he same.

Consumers, businesses, and developers would predictably reject it entirely. But RT’s defeat wasn’t solely on the Office team. RT was also undermined by the shifting dynamics in personal computing, in which the PC, somewhat ironically, was no longer the most personal device people used. It was undermined by Intel’s anticompetitive business practices. It was undermined by hubris, as the Windows team, consumed by its success with Windows 7 and, before that, the Office ribbon, was convinced it could hit another home run. And it was undermined by time. Three years simply wasn’t long enough to orchestrate this many changes.

But it was also undermined by apathy and fatigue. As customers embraced smartphones for personal, entertainment, and casual tasks, they conversely dug in their heels with the PC, a tool that was increasingly just for work. They were no longer interested in innovation or change, but simply wanted familiar tools to keep working. Sinofsky’s teams–Windows and Surface, the latter of which pioneered hybrid tablet PC devices that transformed between different usage “postures”–wanted to “reimagine Windows from the chipset to the experience” to better position Windows against mobile platforms. Microsoft’s customers and partners were far less interested in this shift. The world had changed, was changing in real time as Microsoft moved from Windows 7 to Windows 8 and RT.

When Windows 8 and RT arrived in late 2012, they were met by a stone wall of resistance and even disdain. Sinofsky and the Windows team had underestimated how strongly customers felt about the familiarity of the classic Windows desktop interface, with its Start button, Start menu, Taskbar, and other inconsistent but well-understood interfaces. They were, after all, using traditional PC form factors–laptops and desktops with keyboards and touchpads or mice. And not the hybrid tablet PCs that Windows 8 and RT were optimized for, and that Surface made.

Windows 8 and RT offered identical user interfaces that were simpler and more consistent, with fewer ways to accomplish the same tasks. And they had retained key “muscle memory” actions from previous Windows versions, so that a user mousing down to where the Start button was previously located would find a corresponding interface. But the visceral and negative reactions to their full-screen Start experience and apps surprised the team, as did complaints that Windows 8 and RT had two “modes,” Metro and desktop. Incredibly, it seemed that no one in a leadership role had anticipated these issues.

What they had done was hatch a plan, a new way forward. As the clock ticked down to the Windows 8/RT and Surface launch in late 2012, Sinofsky and his teams worked to solve the problems they had created.

Blue

Speed was critical. They had maintained a consistent three-year development schedule for Windows 7 and 8/RT that was impressive in its rigidity and timeliness, especially with all the work they took on for the newer platforms. But that schedule was glacial compared to the annual updates Apple and Google provided for their more modern mobile platforms. And given the resistance to Windows 8 and RT, each with unique issues, those team would need to improve both quickly while working to further differentiate each from the other. The negativity around Windows RT–a version of Windows that couldn’t even run Windows apps–was particularly acute.

Inspired in part by the long-held belief that Microsoft always required three releases to get any product right, Sinofsky set out to iterate both platforms with annual “rapid release” updates over the next three years. The first of these updates, code-named “Blue,” would address the most pressing complaints–internal, and from partners and customers–with Windows 8 and RT. For the PC-oriented Windows 8, customers would be able to boot directly to the desktop, and the team was looking at bringing back the Start button and, as an option, the Start menu in subsequent releases. But the Arm-based RT was even more problematic, and the ideas that emerged were in some ways contrary to each other.

Windows RT was unique in that it was, and still is, the only version of Windows to ship with a bundled version of Microsoft Office. This wasn’t ideal from Microsoft’s perspective, but because the Office team refused to fully support Metro, all it could do was port the then-current Office apps to Arm and include them in the product. The initial RT release included the desktop versions of Word, Excel, and PowerPoint, plus a Metro version of OneNote. But it lacked Outlook because that bloated app killed performance and battery life–an average of two hours less battery life, Sinofsky later said–making RT even less interesting to businesses. So, Windows RT “Blue” would introduce a paired down version of Outlook that would hopefully eliminate that deployment blocker.

Microsoft also worked to bring file management and other system capabilities that used desktop interfaces into the full-screen Metro world, while adapting both Windows 8 and RT to run well on ever-smaller devices. Where Surface–and, Microsoft hoped, third-party PC makers–would offer hybrid tablet PCs with the initial Windows 8/RT release, the second release would be optimized for so-called mini-tablets. These emerging devices would be used primarily in portrait orientation instead of the more traditional landscape orientation familiar to PC users. And though this change would be available on both platforms, it was particularly well-suited for the device-focused Windows RT and the full-screen Metro apps it could run.

In other words, the teams were hedging their bets, hoping that customers would embrace RT devices, which would drive developers to make more full-screen apps, creating a virtuous cycle that would put RT over the top. Windows 8 would address the needs of traditional PC users, but it would also work well on hybrid tablets and other convertible/2-in-1 PCs. And RT would shift into more of a device platform, similar to the iPad and Android tablets. There would still be overlap between the two–most especially their shared Metro app platform–but each would move a bit further apart too.

Moving on

Sinofsky didn’t stick around to see this plan through. Troubled by the withering criticism of Windows 8/RT that emerged ahead of the launch, much of which was directed at him personally, he suddenly left Microsoft just as the systems launched in late 2012. Rumors claimed he was pushed out, having set his sights on the Microsoft CEO role and angering Steve Ballmer. But whatever the circumstances, Sinofsky left behind a team that struggled to gain its footing in the wake of a growing disaster. That this team was forced to merge with the Windows Phone team–the so-called “B-teamers” they had previously dismissed–was as problematic. Most of Sinofsky’s key lieutenants left the company in his wake, many within the first year.

Nonetheless, the Sinofsky-less Windows team plowed forward with Blue in 2013 while Microsoft’s hardware partners delivered new architectures to address the coming mini-tablet wave. In May, the software giant announced that it would ship Blue as Windows 8.1 and RT 8.1 later that year. And in June, Qualcomm–one of the three silicon makers Microsoft had said would create Arm-based PC chipsets–announced the Snapdragon 800 processor for Windows RT 8.1. This system-on-a-chip (SoC) design featured an integrated 3G/4G LTE modem and USB 3.0 support, and the two companies announced that new devices would appear in the market later that year.

Based on Qualcomm’s phone chips, the Snapdragon 800 seemed ideal for mini-tablets, and Microsoft was (correctly) rumored to be working on a portrait-oriented Surface Mini that would utilize this chipset. It couldn’t happen quickly enough: The original RT-based Surface tablet PC Microsoft released in late 2012 sold so poorly that the company took a $900 million write-down at the close of its fiscal year in mid-2013. And while the second-generation Surface 2 that arrived that fall was a meaningful improvement in some ways, it still suffered from an awkward 16:9 screen with a landscape orientation and its incompatibility with traditional Windows desktop apps.

Microsoft finalized Windows 8.1 and RT 8.1 in October with a restored Start button, the ability to boot directly into the desktop, and other improvements.

Windows RT 8.1 got its stripped down version of Outlook. And there were even a few mini-tablets coming to market, including most notably the Dell Venue 8 Pro, which sold reasonably well and was liked by reviewers. But these mini-tablets were powered by woeful Intel Atom systems-on-a-chip (SoC), not Arm. And Surface Mini was nowhere to be seen.

Blue two and three

Rumors claimed that Microsoft had delayed Surface Mini. But it was, in fact, waiting for the second Blue release, the so-called Windows 8.1 Update. This update would further refine the user interface by displaying icons for Metro apps on the Taskbar on the desktop and offered platform-specific touch- and mouse/keyboard-based optimizations that adapted to the user’s PC. But there were more important updates under the hood, and these were aimed at making Windows 8/RT more appealing to PC makers. Where the initial releases required 2 GB of RAM and 32 GB of storage, the new version would require just 1 GB of RAM and 32 GB of storage.

This could have been a game changer. PC makers could save money by offering lower-cost mini-tablets, passing the savings along to customers. It was just what these cash-strapped companies needed, given their razor-thin margins and the PC sales slump triggered in part by the negative reaction to Windows 8 and RT. It could have likewise been a game changer for Arm. This platform was far more efficient and reliable than the x86 platform championed by Intel, and Windows RT would run far better in the more confined systems.

But once again, Intel stepped in to prevent a competitor from gaining a foothold in the PC market. It heavily subsidized PC makers that exclusively adopted its lackluster Atom instead of the Qualcomm chip. PC makers lined up like lemmings for the payouts and all the mini-tablets that came to market over the next year or two from established players ran the next version of Windows 8.1, not the Arm-based Windows RT. And because this system had lower system requirements, they shipped with underpowered, low performance components that were inadequate for running desktop apps, further undermining the experience and giving Microsoft, and Windows 8.x/RT another black eye.

Sidestepping Intel’s anticompetitive business practices, Microsoft pushed forward with Surface Mini, which utilized the Snapdragon 800 processor and thee 1 GB of RAM and 16 GB of storage minimums allowed by Windows RT 8.1.1. It announced the Windows 8.1 Update–Windows 8.1.1 and Windows RT 8.1.1–in February as the second Blue update in the rapid release era. And new Windows chief Terry Myerson, formerly of the Windows Phone team, previewed new features coming in the third Blue release: Microsoft would bring back the Start menu and let users run Metro apps in floating windows alongside traditional apps on the desktop. It would arrive in late 2014.

But Surface Mini would never come to market.

Strategy shifts

After Steven Sinofsky left in late 2012, Microsoft CEO Steve Ballmer made two major changes. He combined the Windows, Windows Phone, Surface, and Xbox teams into a single organization, with an eye on elevating Windows Phone leadership to drive development of each forward. But he also temporarily allowed two of Sinofsky’s key lieutenants, Julie Larson-Green and Tami Reller, to co-lead this organization to maintain a sense of continuity.

Neither were well-suited to the task, and so Ballmer then assigned the more capable Terry Myerson, who had reported directly to Ballmer since 2011 and led Windows Phone development, to lead this organization instead.

Myerson and his team set out to merge Windows and Windows Phone, and to bring the Metro apps platform to Xbox–and also to emerging new devices like Surface Hub and HoloLens–as part of a “One Windows” strategy. He planned to move forward with Windows RT and Arm for consumer devices like Surface Mini, while sticking to x86 for traditional form factor PCs, including a third-generation Surface Pro. But in February 2014, Microsoft’s board of directors announced that Satya Nadella was the new CEO of the company, replacing Ballmer. And the ruthlessly efficient Nadella had other ideas for Windows and the business unit led by Myerson.

Nadella took the reins at Microsoft at an interesting time.

Steve Ballmer had already set Microsoft down a new path that focused the company extending its leadership in on-premises server products to cloud computing, a massive and fast-growing market for which it was unique poised. But Nadella accelerated and fine-tuned this strategy, making Microsoft a “cloud-first” company in which all its business would need to make sense within the new focus. He marketed this initially as “cloud first, mobile first” as a thematic bridge with the company’s past. But as we would all soon learn, the focus was a lot tighter. Microsoft would meet its customers where they were, and one of his earliest public acts as CEO was to greenlight the Office releases for iPhone and iPad that his predecessor had let languish. Nadella’s Microsoft didn’t care if you used Windows as long as you were using other Microsoft products and services.

In April 2014, Microsoft finalized the $7.8 billion acquisition of Nokia’s devices and services business, which added 32,000 employees and all of Nokia’s worldwide manufacturing facilities. Ballmer had agreed to this acquisition because it was necessary to save the floundering Windows Phone business–Nokia was its only major hardware partner, and it was circling the drain–but Nadella had opposed it. The writing was on the wall.

Nadella met with the senior leadership at the company, asking each to justify the existence of the products they oversaw. Each had to make sense as a standalone business, meaning that it was independently profitable or there was a plan to make it so. And each had to make sense in his new, cloud-focused Microsoft. This was particularly hard on Myerson. Windows was hugely profitable but it made little sense in a cloud company. Windows Phone was failing, and subsuming Nokia would only make that business more expensive to operate. And because it was ignored by its biggest partners–PC makers–and undermined by Intel, Windows RT had never gotten off the ground.

To the mechanically efficient Nadella, the solution seemed obvious. Kill Windows RT, make Windows more efficient by running it like a cloud service, and relegate the Arm hardware platform to non-PC client devices like Windows Phone and HoloLens until it improved enough to make x86 app emulation viable. And if Windows Phone didn’t shape up quickly, and it didn’t, it would have to go, too. (One year later, Nadella would write off $7.6 billion of the Nokia acquisition costs, lay off most of the former Nokia employees, and then kill Windows Phone.)

Myerson’s Surface team was then planning to announce Surface Mini and Surface Pro 3 at an event in mid-2014. They felt Surface Mini had more potential, but both products were major leaps forward. Surface Pro 3 dropped the awkward 16:9 displays used by previous Surface PCs, opting instead for an unusual 3:2 aspect ratio display that proved particularly well-suited for the form factor and remains in use today. It was also thinner, lighter, and more elegant than its predecessors, and it was popular with reviewers and customers alike. Microsoft had, once again, gotten it right on the third release.

When Microsoft held its Surface launch event in June, Surface Mini was nowhere to be seen. Instead, it announced only the Surface Pro 3 alongside new Type Covers and other accessories. Nadella had killed Surface Mini at the last possible moment, with finished units sitting in boxes waiting to be mailed to reviewers. And with that, Windows RT was dead.

Microsoft would announce Windows 10 as the desktop-focused successor to Windows 8.x later that year, but it never mentioned updating Windows RT or a version of Windows 10 that would run on Arm hardware. But when the questions from fans and customers escalated, the company finally agreed to release a final and limited Windows 10-like upgrade to RT. (It was dubbed Windows 8.1 RT Update 3 and it shipped in September 2015, after Windows 10 for x86 PCs.)

It was the end of an era.

Windows RT was dead. Windows 10 would only run on traditional x86 computers, and Windows on Arm entered a Dark Ages of sorts. The dream, it seemed, was over.