Working with partners has always been a big part of the way Microsoft gets its products to people, right back to the start of the company: Altair Basic was a product for another company, and MS-DOS started out as a partnership with IBM.
When Microsoft began building Azure and Office 365, it might have thought that it would switch to selling those directly - and some people, especially consumers and individual developers, do buy direct from Microsoft.
But a lot of businesses still want to work with a partner who can give them some extra help or add more features. With Ignite around the corner, expect a slew of announcements of partner services that integrate with Office 365 - especially Skype for Business - and Azure.
And quite a few of those partners also look like the competition, because who better to help Microsoft integrate Office 365 and Salesforce than, well, Salesforce?
This clearly causes tension sometimes. Okta is a cloud identity management service that covers Office 365, offers single sign on for a wide range of other cloud services and multi-factor authentication, as well as MDM and now API management integration (via Mulesoft and Apigee). That's obviously in competition with Azure Active Directory, and Okta is now Google's preferred identity partner for enterprise customers.
And Box, again, is clearly competition for SharePoint and OneDrive; Box CEO Aaron Levie has eased off on the cracks about SharePoint but even having Microsoft's Peggy Johnson on stage at the Boxworks conference didn't stop him using the Olympic photo of Usain Bolt flying past the other runners to suggest that Box was beating both OneDrive and Dropbox. (Even-handedly, he also asked Google's Diane Greene if Google Plus was one of the billion-user Google services she was touting as suitable for enterprise, reducing most of the audience to laughter.)
Levie asked Johnson about partnerships and competition and her answer is a good summary of how Microsoft decides where to partner, and where to buy or build instead. She said hadn't planned to leave Qualcomm, but she liked the sound of a job that Satya Nadella described as "I want you to make Silicon Valley our new best friend."
She'd expected to be thinking about partnerships the same way she had at Qualcomm - about increasing the total addressable market, getting into new regions and revenue possibilities. "Instead, Satya told me to think about what we can solve for the customer; 'think about the pain points that customers have, put yourself in their shoes'," she said.
Partnering is one way to do that, because, says Johnson, "customers who go to cloud have already made some decisions." If Microsoft can't persuade customers out of those decisions, they're going to partner to support them.
Or as she put it; "Let's make a bigger pie for all of us, not fight over pieces of pie."
I've been saying for a while that Microsoft doesn't buy products unless they're clearly far better than what it could build itself in time, and even then, what it's usually buying is technology and talent rather than the products themselves. It doesn't re-invent the wheel either, though.
Accompli very quickly became Outlook on mobile devices, leaving the team free to change Outlook on the web and the desktop. Wunderlist's recent 24-hour outage might accelerate plans to unify Microsoft's plethora of task management tools; it's unusual that it's survived as a separate product for this long. Yammer is still recognizably Yammer even as it integrates ever more closely with Office 365, but most of the reason Microsoft bought Yammer was to learn to do cloud-style continuous development and delivery and those skills have been behind the arrival of the products built on the Microsoft Graph - from Delve to Groups (which is how Microsoft will compete with Slack rather than buying it).
You can see the same decision making in the way OMS picked up Fluentd to use as its log collection agent for Linux systems, rather than writing its own management agent the way System Center did for Linux and UNIX in days gone by. Not only is OMS using Fluentd, and contributing code; the team has been careful to keep it recognizably Fluentd so that a developer could take the Microsoft packaging of Fluentd and use it as Fluentd without ever using OMS.
Similarly, I've argued that Microsoft is unlikely to buy, say, GitHub, because they can get everything they want by partnering with them. Thinking about whether a partnership makes for a bigger pie is an easy way to tell if Microsoft is likely to make the partnership, however odd it might once have seemed.