The price of autonomy in IT: what MSPs have to choose now

The price of autonomy in IT is rarely put on the quote. You only notice it when a client calls to ask whether they can get off Microsoft, and within ten minutes you realise the answer is “yes, but not for the next five years”. That is the real conversation on the table for MSP owners right now. And it is a conversation most owners have not prepared for, because for years nobody asked the question.

Research by Proximus Nexus among end users shows that around 68% of organisations see sovereignty as a non-subject. Only twenty per cent believe they are not sovereign with their data. In practice that means most of your clients are not asking an active question. They run on Azure, AWS or Google, that is allowed, so that is what they do. And you keep reselling.

The problem is not that hyperscalers are bad. They are technically excellent, they offer portability between regions, they save your client when the data centre in the Middle East is on fire. The problem is that the question of when your client wants out is going to come more often and more forcefully. From governments first, then from banks, then from mid-sized business. And if you as an MSP are then only a reseller of one stack, you stand there empty-handed the moment your client board decides the risk profile has changed.

Why the price of autonomy in IT is on the table now

What works is not: pulling your entire client portfolio off Microsoft in one weekend. That is not going to happen. What does work is offering the client two tracks. Track one: leave the existing email, the existing stack, the existing Office environment in place. Change as little as possible there. Clients need calm around what works.

Track two: use the new demand, and that demand is now almost always AI, to bring in a sovereign alternative alongside it. So not selling Copilot and pulling the client deeper into Microsoft. Instead running an AI layer you host yourself, based on several models, and integrating it with the offer you already have. That keeps the client productive, and step by step you build a second leg that does not stand in Redmond. That is the only realistic route where you can show something within 12 months.

Hotel California: why getting out really hurts

An owner asked it straight out recently: are we in Hotel California? The answer is yes. You could get out, but not for now, and it hurts. At one client we support, a decision of principle has been taken to move off Microsoft towards an open source European alternative. They accept a loss of functionality. They accept that some people are literally cut off from tools they were used to. That is what sovereignty costs at the moment.

From governments the signal now comes more and more often: we are willing to pay more for less functionality. That is a sentence nobody spoke three years ago. Now they do. And that is exactly where the business question sits that MSP owners have to understand: it is no longer the IT department that solves this. It is the board that sets the risk profile, and that asks you what options you have. If all you can answer is “Azure or AWS”, you are out of the conversation.

What this means in concrete terms for the average MSP: over the next two years you will be having conversations in which the client no longer asks “what is the cheapest option”, but “who has access to this data, and under which jurisdiction”. That is a fundamentally different conversation. Price is then no longer the first variable. Legal basis is. And the MSP with no answer to that gets replaced by the MSP that does have one, even if it is more expensive.

The portability problem nobody talks about

What you have seen happen a lot in the Netherlands: DevOps teams allowed to make their own choices, all sorts of enjoyable hyperscaler services added, and years later nobody is able to move that workload. A cloud strategy does not start with choosing a provider. It starts with an architecture that is portable. Kubernetes, open source components, technology that makes it possible to shift between AWS, Google and Microsoft, and before long between a hyperscaler and a European sovereign cloud. If the answer to “can anyone from outside reach this data” is yes, you are by definition not sovereign, however high your average compliance score may otherwise be.

For MSPs that do not develop technology themselves, which is the great majority, this means one thing: you have to start building a second reseller line now. European players are coming, partly because government is slowly rewriting its tenders. As long as tenders are still drawn up with conditions that in practice only hyperscalers can meet, smaller European parties stay on the sidelines. That will shift. And on the day it shifts, you do not want to be the one still working out which European supplier fits which client profile. By then you want those conversations three quarters behind you.

What you actually do on Monday morning

Enough theory. What does this mean for next week diary? Three things. First: take your top twenty clients and put them in a simple matrix. On one axis: how deep they sit in the hyperscaler stack, measured in the number of integrated services. On the other: how likely it is that their sector will face sovereignty requirements within two years. Healthcare, government, financial, defence supply, critical infrastructure, those are at the top. That list is your priority for the conversations you want to have in Q1 and Q2.

Second: start with one pilot, not ten. Choose one client, preferably a mid-sized one with a motivated IT lead, and build a concrete use case there in which you put up a sovereign AI layer or a portable workload component. Not as a replacement, but next to what is already there. The aim is not to migrate that client completely. The aim is that you learn for yourself what goes wrong, which integrations hurt, and which conversations you need to have with a supplier before you can offer it to ten other clients. Without that learning experience you will keep parroting marketing brochures.

Third: once a quarter, have a conversation with your three largest clients in which the words “exit strategy” simply come up. Not as a threat, not as a sales pitch. Just: if you wanted to switch in three years, what would that look like? Simply by asking the question you position yourself as the party that sees the wider picture. And you get back information you did not know the board was working on. That is why MSPs doing this now will still be the board conversation partner in two years, while others are pushed back to the level of supplier.

The price of autonomy in IT is that you have to work on it now, while the revenue still comes from the old stack. Anyone who waits for regulation or for a large client to force it ends up in a position where the choice is made for them. Anyone who builds a second leg now, however small, even if it is one client with one AI use case, will be on the right side of that conversation with the board.