SAP and Dynamics in an AI world: why closed ERP rarely fits the mid-market

At the end of 2027, SAP stops mainstream maintenance of ECC. Anyone staying on Business Suite 7 then buys extended support until the end of 2030 and migrates after all. You did not choose that date. SAP did.

For a group with thirty people in its SAP department, that is a line in the multi-year budget. For a company of two hundred, it is a rebuild of the system that runs invoicing and accounting, precisely when you want to move forward with your data and with AI. SAP and Microsoft make solid software, that is not the issue. The question is whether a closed suite with a per-user price fits a business that wants to move quickly and own its own data.

Indirect use, direct invoice

Take SAP v Diageo. The drinks group had properly licensed SAP for its own staff, but attached an application on Salesforce through which customers and salespeople entered orders themselves. According to SAP, that was indirect use of the SAP software, separately subject to licensing. The bill rose to more than 54 million pounds, and the High Court ruled in SAP's favour on liability in 2017.

The sting is in the model. You link a web shop, a customer portal or an AI assistant to your ERP, and afterwards every indirect reader counts as a paid user. Just when you want to connect systems to do something with that data, it becomes a brake you only feel when the invoice arrives.

The price is tied to users, not to results

Microsoft Dynamics 365 Finance sits at 210 dollars per user per month, the Premium variant at 300. Straightforward, until you work it through. A hundred people who touch the system cost well over a quarter of a million dollars per year in licences alone, before anything is built. The implementation itself runs, according to market figures, from 250,000 to more than 1.5 million dollars.

That model has a logic that works against you. Growth costs users, and users cost money. Add ten people on the shop floor and your software invoice rises with them, even though that tenth user brings in no extra euro. You pay for access, not for outcome.

That touches on exactly what matters now. Anyone who wants to do something with AI starts with data they are free to access and move. In a closed suite, your data sits in the vendor's tables and format, and every link to a model or a dashboard risks precisely the discussion about indirect use for which SAP claimed more than 54 million pounds from Diageo. Ownership of data is no longer a matter of principle. It is the condition for keeping AI affordable.

The larger the project, the larger the deviation

Then the migration itself. McKinsey, together with the University of Oxford, examined large IT projects of more than 15 million dollars. The picture is sobering.

That 17% deserves a moment. One in six large IT projects threatens the survival of the business that starts it. A large group has reserves to absorb such a blow. A mid-market company does not, and with a big-bang migration you bring that risk on yourself.

Where SAP and Dynamics do fit

This is not a case against SAP or Microsoft. For a multinational with locations in thirty countries, hundreds of legal entities and consolidation across a handful of currencies, those suites do work that little else can handle. There the scale justifies the complexity, and the weight of the vendor is reassurance rather than risk. It is about fit. What a group of ten thousand people carries will flatten a company of two hundred.

Ownership instead of subscription

For the mid-market, there is another way. Build on an open-source foundation that you own yourself, with the code in your own repository from day one. Growth then costs infrastructure instead of licences per head, and a linked AI application or web shop is simply an integration, not an after-charge for indirect use. Our platform runs on a standard stack with Kubernetes, PostgreSQL, ClickHouse and Metabase, so that your data stays yours and remains transferable.

And it does not have to happen in one go. Our approach works in waves with a go or no-go at the end of each step, so that you never take on a project of more than a year without being able to stop along the way. If you first want to know where your current stack leaks in cost, time or security, then an audit exposes that before you replace anything.

Today's ERP choice determines how freely you work with your own data over the next ten years. A closed suite with a deadline you did not choose and a price that grows with every user is rarely the answer for the mid-market. Ownership, an open stack and the right to stop count more heavily the faster you want to move forward.