The cost of building custom software is falling as AI tools accelerate development. At the same time, vendors are embedding AI into packaged applications at speed. The result is that enterprises can no longer apply one sourcing strategy across the board. Each workload demands its own decision: build in-house, buy from a vendor, or partner with a specialist.
For SAP-centric organisations in the DACH region, this creates a staffing challenge. The people who can evaluate these trade-offs, workload by workload, need to understand lifetime cost modelling, data residency implications, and vendor lock-in risk. A client in Frankfurt recently restructured a senior architect role mid-search because they needed someone who could own the sourcing framework and present it to the board. The functional and technical skills remain, but the commercial reasoning has moved up the requirement list.
Data control is the dimension that keeps surfacing. When AI is involved, companies must decide where training data lives and who can access it. Organisations with strict data residency requirements, common in regulated industries across Germany and Switzerland, often find that building internally is the only option that satisfies their compliance teams. Others decide the vendor route is faster and cheaper, provided they can negotiate the right contractual protections.
The risk calculus matters too. Building custom AI capability sounds appealing until a company realises it has created a dependency on a small internal team. Partnering spreads that risk but introduces coordination overhead. Buying shifts the risk to the vendor but limits flexibility.
Companies are starting to ask for candidates who can hold all three options in view simultaneously. The ability to frame a sourcing recommendation and defend it to a CFO is proving as valuable as deep technical knowledge in S/4HANA or data architecture.
Prompted by reporting from ERP Today.