View of Microsoft building showing cloud computing licensing

Licensing or Lock-In? Evaluating Microsoft’s ‘New’ Approach to Cloud Competition

About five months ago, Microsoft President Brad Smith promised publicly to update licensing restrictions that since at least 2019 had the effect of raising costs for the use of Microsoft enterprise software on competing cloud services other than Azure. I wrote at the time to express skepticism about the promised changes, which in my view seemed heavier on the PR dimension than on significant market reform. We now know more about Microsoft’s plans—and I remain skeptical. Microsoft needs to accept a level playing field for competition in the cloud, and not revert to the bad old days of tying, bundling and other gimmicks to boost the market share of Azure through anti-competitive practices.

Microsoft in May promised its European partners a simplified licensing regime that would be easier to understand, and by implication that regulators could more easily assess its impact on the competitive environment. It was a welcome promise, but it simply hasn’t been delivered. With respect for some of the necessary complications that software licensing regimes can involve, the new proposals are convoluted and at the same time vague on important details that are material to their impact. For example, consider the proposed reform to the Services Partner Licensing Agreement (SPLA):

At its inception, SPLA was intended to allow partners to offer hosted services from their own datacenters, not for managed service providers buying through SPLA to host on others’ datacenters. We are making changes to the SPLA program, starting in October 2022, to better align with the program’s intent, and with other commercial licensing programs. To strengthen the hoster ecosystem by focusing the program on breadth hosters and encourage traditional outsourcers and datacenter providers, we are changing the SPLA terms to remove the ability to outsource SPLA licenses on Listed Provider datacenters.

Buried in this language is, in fact, a simple message that could and should have been delivered directly: Microsoft is erecting new barriers for partners who want to deploy Microsoft products on certain other cloud providers, like AWS or Google Cloud. And the supposed justification for this—that it would somehow affirm the original intention of the SPLA program—is besides the point. The technology world and in particular the cloud is a very different place now than it was when Microsoft launched SPLA. And that Microsoft would seemingly prefer a smaller number of large SPLA users who run their own data centers rather than a larger number of smaller SPLA users who would more likely run on top of a cloud infrastructure, is certainly not what makes sense in today’s cloud environment. It is bad news for the cloud ecosystem and for Microsoft’s partners who want to operate in a more modern and efficient way. If Microsoft believes otherwise, it needs to explain why.

Brazenly, Microsoft’s plan appears to introduce a number of new contingencies and restrictions that sharply limit possible benefits. For example, the ability to move a current software license from an on-premise server to the cloud seems to be contingent on the customer joining a Microsoft software assurance program. This restriction will make what sounds like a desirable move toward a BYOL (Bring your own license) regime more like a backdoor means of increasing customer dependence on Microsoft. This and other provisions are most likely to have the effect of inserting Microsoft into the middle of a current relationship between the end-user and the cloud-based services provider, without any justification for why that provides value to anyone other than Microsoft.

But by far the most important problem with Microsoft’s proposed changes is the inexplicable exclusion from the program of what Microsoft calls “listed providers.” What does this mean? In 2019, MSFT created out of thin air this special category, which now includes Alibaba, Amazon Web Services, Google, and in a peculiar twist, Microsoft itself. (This is a bit of bait-and-switch explained more below.)

Microsoft has never specified what criteria would land a company on this list or explained why the list even exists from a technology, economics, or business perspective. Why should listed providers be subject to different licensing provisions than anyone else? How can a small European cloud services provider, or any other supposed beneficiary of the regime change, be confident that they won’t wake up one morning and find that Microsoft has added their name to this list?

One can intuit answers to these questions that are not particularly friendly, but it should not be the job of outsiders to figure it out. The burden rests on Microsoft to explain clearly why the main potential benefits of its new licensing regime should be available to smaller cloud providers but not to “listed providers” and, by implication, not to customers who wish to use those listed providers. This burden becomes even more important because Microsoft includes provisions for users of Azure to “escape” from the restrictions on listed providers, which makes the inclusion of Microsoft on its own list seem particularly cynical.

Why does this matter so much? Because, as I’ve argued before, competitive markets are not places where you get to choose who you compete with. That is for the market to sort out. No firm should ever be in a position to determine on its own that full stack competition is not permitted by virtue of its unilateral practices.

Microsoft appears to be erecting new barriers for partners who want to deploy Microsoft products on certain other #cloud providers, like AWS or Google Cloud Services in proposed reform to their Partner Licensing Agreement (SPLA). #respectdataClick to Post

Keep in mind that the promise of cloud computing is much more than just relieving users from having to run their own on-premise hardware. Cloud enables something more profound and transformative—truly elastic computing, where compute is treated as digital work that gets distributed without friction to be done where, when, and how it can be done most effectively and efficiently, regardless of the level of the stack. In other words, elastic compute includes applications as well as physical and logical infrastructure. It includes the whole stack. That’s what cloud technology is really about, and the promise is too important to the economy as a whole to let it be hijacked by the anti-competitive practices of any firm.