Author: Peter M. Färbinger, E3 Magazine

To understand the immense importance of MCP, one must consider the structural integration challenges of modern ERP landscapes. Until now, integrating AI with existing S/4 systems has been a tedious and proprietary nightmare.

Now the decision has been made, and Anthropic has had to put two of its most powerful LLMs on hold. Just a few weeks ago, SAP CEO Christian Klein stated that Anthropic would be a key technology pillar of SAP’s AI strategy.

SAP’s Autonomous Enterprise offers a wide range of opportunities, from Agentic AI to Composable ERP. All the necessary IT tools for this are available in SAP’s portfolio. Ultimately, however, the question of affordability—based on licenses, maintenance, and Digital Access—remains.

Tabular foundation models seem tailor-made for SAP’s ABAP world. In an ERP system where everything is based on tables, an LLM (Large Language Model) has no place. With TabPFN (Tabular Prior-data Fitted Network), the startup Prior Labs can save SAP’s Autonomous Enterprise.

Naturally, in the first half of 2026, there was criticism of our criticism of Christian Klein and his fellow board members. The E3 editorial team is concerned and is pointing out the problems in a critical yet constructive manner. The result is truths that SAP, the global ERP company, does not like to hear.

AI knows only one path: to solve a given task optimally and once and for all! What seems like a radical approach to humans is, for the machine, its logical raison d’être. Because AI has no humanistic values, it will continue to optimize until its very end.

With an antitrust ruling, the EU has provisionally brought an end to an anti-competitive chapter in SAP’s history. The global leader in ERP was accused of exploiting its dominant market position in maintenance and support to force customers into an inescapable service straitjacket.

The cloud exit problem has not yet been solved, and already the need for another exit strategy for AI is becoming apparent. How will sovereignty ever be achieved?

Subscription or consumption? Buy, rent, or pay based on usage? Following the on-premise era, existing SAP customers are facing an almost insurmountable complexity of fees. The Business Data Complexity (SAP BDC) defined by the user association DSAG is just the tip of the iceberg.

First, my colleague Christof Kerkmann reported in the *Handelsblatt* on the restructuring of SAP’s Executive Board: CEO Christian Klein plans to focus on AI technology going forward, as Executive Board member Muhammad Alam (Product and Engineering) will be leaving the company in early 2027. Klein is supported behind the scenes by a strategic advisor.

"The Beginning of Better": Christian Klein expressed his hope for an improvement in SAP's difficult situation on the Sapphire stage in Orlando.

SAP is leading its existing customers on a journey of its own accord (autonomously) while they strive to reach the summit with their ERP system.