Microsoft’s Symbolic Shift from Windows to Cloud

I’m attending Microsoft’s Build developer conference this week and one of the most striking things is the way in which Microsoft has reversed the order of its daily keynotes from past years. Whereas the first-day keynote in the past has focused on Windows and devices, those topics have now been pushed to day two, while cloud and AI dominated Wednesday’s first-day keynote. That’s an administrative shift but an important one symbolically. It represents the broader shift in Microsoft’s business away from its legacy focus on Windows and Office software and towards a cloud- and services-based future.

From Mobile and Cloud First to Intelligent Cloud and Edge

When Satya Nadella took over as CEO of Microsoft, he replaced Steve Ballmer’s Devices and Services mantra with a focus on “mobile first, cloud first“. That was an important strategic shift for a company which had invested, under Ballmer, in building and buying device businesses, and was on the cusp of a big shift to services, but which still largely favored its own devices and platforms over a platform-agnostic approach to those services. Nadella’s new focus said the first priority was making sure Microsoft’s products and services were optimized for mobile (most of which was dominated by Google and Apple from an operating system perspective) and cloud (where Amazon and other companies played major roles and Microsoft’s business was again a minority of the market). That enabled the more aggressive expansion of the Office family to third party devices like iOS and Android, more open sourcing of functionality, the expansion of Azure and AI across platforms, and so on.

Now Nadella is pursuing a new vision, which is about Intelligent Cloud paired with Intelligent Edge. That’s a little less intuitive than the old mobile- and cloud-first vision, so it’s worth exploring a bit. There are several components to this shift. First, mobile is gone, not because it’s no longer relevant but because it’s no longer distinct from the market as a whole: building for mobile is now a given and no longer needs to be explicitly called out, at least in theory. Second, the focus on cloud remains and now expands to the totality of the vision but the cloud component now has two parts to it. First is the classic cloud core – all the servers in data centers around the world, with heavy infrastructure driving massive processing and computer power but the second part is the new and important element. Doing cloud-style processing using traditionally cloud-based processing in edge devices, whether that’s PCs, mobile devices, or tiny low-powered IoT devices like Raspberry Pi.

The benefit of that approach is that “cloud” tasks now no longer need to take place remotely from the devices and services they serve, which can reduce latency, reduce the amount of data that needs to traverse networks, and enable businesses and their processes to be nimbler. On stage at Build, Microsoft demoed an example of this approach centered on a Scandinavian manufacturing machinery company. But it has lots of applications which will start in industrial settings but could eventually spread to other enterprise and even consumer scenarios.

That’s a somewhat unique vision for cloud, with many competitors still largely focused on the large core applications and putting more power into data centers. Apple and Google have, to some extent, been experimenting with doing more machine learning and AI tasks on device in the past year but they’re doing that mostly in their device and OS businesses rather than their cloud businesses. In other words, those efforts are largely separate from their cloud businesses rather than integrated into them. The promise of what Microsoft is doing is a consistent and integrated cloud business that seamlessly works across core and edge elements. To be clear, that’s still the theory rather than the practice at this point even for Microsoft, but it’s a distinct vision.

Cloud as an Enabler of AI

One of the phrases from the analyst meetings Microsoft held ahead of Build earlier this week that stood out to me the most was “Cloud is a warmup act for AI”. The point was that cloud, together with other big architectural and structural shifts in IT, have been enormous and have enabled many things but perhaps most importantly, they have enabled AI to flourish and become truly useful and critical to future innovations. So AI itself was a key second theme from the first-day keynote at Build and one which had some important themes in its own right. Among those were:

• Democratizing AI – although I don’t think I heard this phrase uttered explicitly on stage, it’s been a consistent theme for Satya Nadella lately. What it really represents is the enablement of AI and machine learning for developers in such a way they don’t need their own expertise in the field to make use of it. That definitely was a theme in the keynote and continues to be an important element of what Microsoft is doing with AI beyond its own use in first-party products like Office
• Custom Features – another theme of the keynote was enabling customization of image and text recognition for developers so they can train AI and machine learning with their own unique and obscure data sets, again in such a way they don’t need to understand the tasks themselves
• Automation and Cognitive Skills – one last theme I’ll mention is the combination of steady improvement in core cognitive skills like image and text recognition, language processing, translation, and so on. Microsoft is getting very good at each of these individually but it’s also getting better at putting them together and then providing automated functions like real-time translation, something Microsoft demoed on stage in impressive ways.

Microsoft is certainly one of the leaders currently in AI and especially in the field of enabling generalist developers with AI and machine learning capabilities. I’m always wary of declaring winners in such a complex and broad field but it’s certainly up there with other broad-based leaders like Google and Facebook, while still others like Apple use AI more quietly and in more focused ways to improve their own products.

More to come on Windows and Devices

Microsoft, like Facebook and Google, has two keynotes during its developer events, which is indicative of just how broad its business is at this point and, although Windows and devices have been pushed to day two this year, there will be plenty to talk about in terms of moving Windows 10 forward through the next update later this year, Windows Mixed Reality and HoloLens, and so on. Microsoft isn’t leaving this stuff behind by any stretch of the imagination. But it’s the cloud, AI, and this other cross-platform stuff that’s driving the majority of Microsoft’s growth today, while Windows and devices represent a more flat to declining business financially. Cloud and AI are the future and they’ll enable new functions and features of Windows and devices as we’ll see in the day-two keynote too. The order of those things is perhaps the biggest shift in Microsoft’s business and we’re really starting to see that play out now.