In February 2026, DSAG, the German-speaking SAP User Group, published its annual Investment Report. Among the 198 member companies surveyed across Germany, Austria and Switzerland, 43% said they had put AI use cases in place. Of those, 77% were running them in production on non-SAP tools, and 3% on SAP.

SAP reports more than 34,000 of its 320,000 plus customers running AI in daily operations. That’s roughly 1 in 10. Both figures are accurate, and they describe very different pictures of the same install base. That’s the argument this article is about.

SAP keeps announcing things. At Sapphire 2026, it launched the Autonomous Enterprise, a new Business AI Platform, and an Autonomous Suite with 224 agents and 51 assistants. In the same window, 62% of DSAG members said SAP’s Business Suite target vision guides their investment planning little or not at all.

So we asked the community what they were seeing. Mentors, independent advisors, and experienced practitioners who spend their days inside ECC and S/4HANA and have watched enough technology cycles to tell a wave from a tide.

Key takeaways

• The DSAG Investment Report 2026 found 43% of surveyed member companies had implemented AI use cases. Among those that had, 77% ran them in production on non-SAP tools and 3% on SAP. Sample: 198 DACH companies, December 2025 to January 2026, with public administration more heavily represented than in past years.
• SAP reports more than 34,000 of its 320,000-plus customers running AI in daily operations, roughly 1 in 10. Both numbers are accurate, and they measure different things.
• Intent is climbing even where production isn’t. DSAG members planning high or medium BTP investment specifically in AI went from 2% in 2024 to 16% in 2026.
• The barriers practitioners name: unpredictable usage-based pricing, governance and security, trust in the output, and people and processes moving slower than the tools.
• Several point to a regulatory overhang as well. Some leaders held off on SAP AI decisions pending the European Commission’s ERP support case, which closed on 9 July 2026 with SAP’s commitments made binding for ten years.
• The AI practitioners trust most is the oldest: predictive analytics, document grounding, OCR.
• They split on where the bottleneck sits. Some say SAP’s architecture, licensing and cloud requirements. Others say organizational readiness. A third group says the technology is ready and clients are waiting for proof. That split is the most honest thing in this piece.