AI Integration Gaps Are Forcing the Clean Core Debate into the Boardroom

Hiroshi Ozaki connects SAP’s operating signals to executive decisions

I have spent 35 years watching enterprises add layer upon layer to their core ERP systems. First came bolt-on supply chain modules, then customer experience platforms, then a wave of point solutions during the early cloud years. Each addition promised agility, but too often it simply buried the single source of truth under a maze of interfaces. What we are witnessing today with artificial intelligence is different. AI does not just add a new layer—it shines a floodlight on every crack, every data inconsistency, and every half-finished integration that has accumulated during decades of tactical IT decisions. For executives and investors, this is no longer an architecture debate. It is a balance-sheet risk, and a leading indicator of who will capture value from the next wave of enterprise software spending.

The Business Signal

The financial community has been tracking SAP’s cloud backlog and S/4HANA adoption rates as primary gauges of the company’s transition. But the AI integration gap adds a new dimension. As the original analysis highlights, AI pilots are consistently exposing silos across ERP, supply chain, and data domains that had been papered over by human workarounds. When an AI agent tries to autonomously adjust inventory based on a demand signal, it fails if the demand forecast sits in a separate cloud application, the inventory master data is not harmonized, and the security model was never designed for machine-to-machine transactions.

For SAP, this creates a commercial moment. The company has been leading with the message of a “clean core” for years, but many customers saw S/4HANA migration as an infrastructure refresh—a necessary cost, not a strategic unlock. AI now reframes that investment. A fragmented landscape cannot support trustworthy autonomous decisions. So every board that is under pressure to deploy generative AI in supply chain or finance is suddenly forced to fund the integration work that was deferred. This has direct implications for SAP’s current cloud backlog and partner deployment pipelines. If enterprises get serious about AI at scale, the remediation pipeline for integration and master data governance could become the next significant driver of S/4HANA conversions and RISE with SAP subscriptions. Conversely, if companies attempt shortcuts—wrapping AI wrappers around silos—the inevitable project failures will create headline risk that slows the cycle.

What to watch: SAP’s quarterly cloud backlog growth rate and S/4HANA adoption numbers will, over the next two to three years, be influenced by how many large enterprises link their AI roadmap to core ERP modernization. Listen for management commentary connecting AI wins to clean core adoption; it is a sign that the integration gap is being monetized.

What It Means for SAP Customers

For transformation leaders inside large enterprises, the message is uncomfortable but clear. AI is not a standalone project you can assign to a data science team. It is an architectural stress test of your operating model. In my consulting practice, I see three immediate priorities for customers:

1. Audit and simplify before scaling. The article’s call to audit ERP, supply chain, and data systems is not optional. I have seen too many organizations launch an AI pilot with “representative” data only to discover that the production environment has twelve different definitions of a customer record. AI amplifies that disconnect into wrong orders, compliance violations, or worse. A pragmatic first step is to map the integration flow for a single end-to-end business process—say, order-to-cash—before training any model.

2. Establish data governance as a business discipline. The phrase “trusted context” from the source analysis is precise. AI consumes context; it does not create it. If your data lacks provenance, quality, and security lineage, every AI output becomes a liability. This is not an IT problem. It means the CFO’s organization must own data definitions that span finance, supply chain, and procurement, and the CHRO must align incentives across functions. This cultural alignment, which I have long emphasized, is the hardest part.

3. Use AI pilots as diagnostic tools, not for external bragging rights. Let the first pilot fail gracefully in a controlled setting, and catalog exactly where it broke. Was it a master data mismatch? A security token expiry? A supply chain visibility gap? The resulting “gap map” becomes the business case for remediation—funded not as theoretical technical debt reduction but as a prerequisite for future AI capabilities. This approach turns an embarrassing failure into a strategic investment roadmap.

The counterargument: some executives may conclude that AI is overhyped and postpone action. In my view, that is risky. Generative AI is maturing faster than the cloud wave did, and competitors who use AI to compress planning cycles or automate exception handling will widen operational performance gaps. Waiting until your integration issues cause a production failure in an AI-driven process could be far more expensive than the upfront investment in clean core and data governance.

What Market Observers Should Watch

I propose three leading indicators that will separate companies positioned for AI success from those drifting toward a costly reset:

• RISE with SAP adoption among late-majority S/4HANA customers. If a significant cohort of manufacturing and automotive companies suddenly accelerates their move, it may be driven less by the 2027 end-of-maintenance deadline and more by the realization that AI can only be deployed reliably on a harmonized backend. Watch for large deals that bundle business technology platform (BTP) and integration services—those are signs of AI-driven clean core demand.

• Partner ecosystem capacity for integration and data governance work. The systems integrators who spent the last decade building cloud connectors now face a different kind of complexity: cleaning up on-premise custom code and reconciling semantic layers. If the major SI firms report a backlog of “AI readiness assessments” that convert into multi-year transformation contracts, that is a tangible signal that the integration gap is being addressed at scale.

• Regulatory momentum around AI and supply chain transparency. The EU AI Act and emerging supply chain due diligence regulations are forcing enterprises to prove they can trace decisions. If regulators begin to probe the data provenance behind automated supply chain or pricing decisions, the cost of fragmented landscapes will jump immediately. This is a catalyst that could compress the timeline for remediation from years to quarters.

The caveat: SAP’s own AI narrative must not outpace customer reality. If SAP’s marketing overpromises on AI agents that “just work” across hybrid landscapes, it risks the same credibility gap that slowed early cloud adoption.

References

• AI Reveals Vulnerabilities in the Enterprise Operating Model
• AI Reveals Vulnerabilities in the Enterprise Operating Model
• SAP Security Notes & News

References

• AI Reveals Vulnerabilities in the Enterprise Operating Model
• AI Reveals Vulnerabilities in the Enterprise Operating Model
• SAP Security Notes & News