Has Microsoft overstretched its cloud elasticity?

Ask any technology decision maker of the last decade why they want to move to public cloud and elastic scalability will likely be in the top three reasons. That’s because of the promise of effectively infinite available capacity on demand, for as long as you might need it.

It doesn’t actually matter if new customers make use of that scalability, and in fact most of them have probably not done so (especially if they simply lift-and-shifted non cloud-native apps to the cloud). The key decision point was that the capacity was there, ready and waiting at your command.

However, there are growing signs that – for Microsoft at least – this open-ended supply of cloud resource might indeed be limited, and potentially even restricted in some markets and sectors. 

The most recent sign of this are reports that instead of continually growing out their planned Azure platform for GitHub, Microsoft have been using AWS to bolster their capacity and improve GitHub’s resilience, which has suffered multiple outages since Redmond took it over.

GitHub has cited multiple issues around tight coupling of services to the underlying infrastructure as well as simple capacity constraints. These are issues that, arguably, transfer directly across to normal cloud platform users and are often accused of driving vendor lock-in.

Other areas of Microsoft’s empire also show signs of scaling challenges. These include their roll back on November commitments to deploy fully Sovereign AI inferencing. That’s now down to just four of their cloud product areas, with delays of up to two years for Japan, whilst generalising specific national promises to Germany, Italy, Spain,  Sweden, Poland and Switzerland, who will now have to share an EU and EFTA regional service.

Taken together, it is fair to wonder if Microsoft has – as has been the case before – simply overpromised on what it can deliver? 

The signs that it has are hard to ignore.

If so, the obvious customer response might be to broaden cloud landing options and diversify into a multi-cloud model – though Microsoft’s decision to shed some resources over to AWS for GitHub suggests you might already be multi-cloud but not fully realise it.

The positions of the UK CMA who are investigating possible Strategic Market Status (SMS) and EU regulators who appear close to applying gatekeeper status to Microsoft and AWS, are that you ought to be able to do so – but saying that is a lot different to being able to actually do so.

Moving from one cloud platform to another remains a difficult task – this isn’t like having an open ticket and stepping off an overloaded train to an adjacent service run by another operator. Cloud vendor lock-in is a very real thing, and far too many of us today ride on a restricted “this service only” cloud ticket.

Technical issues that prevent simple transfers also start well away from the cloud itself. At the desktop, the corporate identity you rely upon for laptop login is often the same one that governs all your cloud services and isn’t directly and openly interchangeable with other cloud providers. Whoever holds your digital user identity is immediately in a preferential position.

This is why the CMA investigation under the Digital Markets, Competition and Consumers Act (DMCCA) is so important and complex. Cloud selection and mobility is more often than not rooted in internal desktop, identity and server room infrastructure and that gives Microsoft a heavy bias.

The pressure to move between clouds and mix and match to best commercial and technical fit has never been greater, but whether the CMA and EU regulators can find a way to move us from “this service only” to “use any cloud” nirvana remains unclear.

They have the levers to do so, but need to examine the full stack effect and consider whether moving between clouds – while also considering potential scaling limitations from Microsoft – are sufficient reasons to open the market up in the interests of the end user.