“Open” on Whose Terms? What SAP Sapphire Won't Say.

SAP Sapphire kicks off in Orlando this week. Christian Klein will take the stage and tell you enterprise AI is finally real. That the data problem is solved. That SAP is building an open, intelligent platform for the future.

But Sapphire is not where SAP's strategy gets made. It's where SAP's strategy gets announced. To understand what they're actually building, look at what they do before they take the stage, and what they do after the cameras stop rolling.

Silent Policy Speaks Volumes

In late April, SAP updated its API policy. Section 2.2.2 now explicitly prohibits the use of SAP APIs for interaction or integration with autonomous or generative AI systems that plan, select, or execute sequences of API calls, unless that happens through SAP-controlled architectures and pathways. It also prohibits large-scale data extraction or replication outside SAP-sanctioned routes.

SAP is also taking steps to prohibit and technically block data access by third party systems using the ODP API, potentially breaking existing connections.

(Sidenote: If you’re concerned about the new API policy, you’re not alone. SAP has a lengthy FAQ that’s 17 pages long to address the many issues.)

In plain English: if you want AI agents to work with your SAP data, SAP wants that to happen on SAP's terms.

The community reacted immediately. Consultants called it lock-in. Customers panicked due to uncertainty. The German SAP user group DSAG issued a sharp public criticism. They raised concerns about the scope and implications of the updated policy, specifically that it threatens customers' ability to adopt AI innovations that connect to SAP systems, on their own terms.

Klein walked it back on the Q1 earnings call. Reassured everyone SAP wants to be open. But the language in the policy itself hasn't materially softened. Legal rights and practical access architecture are two different things.

This is what SAP does away from the keynote stage. Note it. Because after Sapphire, when the announcements have landed and the coverage has moved on, the policy will still be there.

A Strategic Pattern Emerges

The API policy did not arrive in isolation. In the same period, three acquisitions in quick succession: Dremio, Prior Labs, Reltio. Acquisitions are complex and take time to integrate. In the meantime, they can create uncertainty for customers around roadmap direction, pace of innovation, and long-term product strategy.

• Dremio was one of the genuinely open data lakehouse platforms on the market, built to give organisations high-performance access to data wherever it lives, without forcing it into any single vendor's stack. That independence was the point. Now it will belong to SAP.

• Reltio is master data management. MDM defines the canonical version of your most critical enterprise data, customers, suppliers, products. It defines what's true. SAP's stated view is that its ERP is "the institutional brain of every company where data and process domain know-how is stored." Owning your MDM layer is not just storing that knowledge. It is becoming the authority on it.

• Prior Labs brings AI that understands structured enterprise data natively, trained to operate within SAP's ecosystem. The more it learns about your data, the less portable that intelligence becomes.

Taken together with the API policy, a pattern emerges. SAP is not building openness. SAP is buying the exits and locking the doors.

The Trade-offs Customers Should Be Questioning

Whether you are in Orlando for SAP Sapphire or watching the announcements from afar, the real questions may not be the ones on the keynote slides. They are the questions customers need to ask before they commit more of their data and AI strategy to a single vendor-defined architecture.

• How much flexibility are you giving up? If your data can only move through approved pathways, your ability to choose the right cloud, platform, or AI service becomes constrained by someone else’s roadmap.

• Where do the costs really show up? AI changes the economics of data. When data access, replication, storage, and external compute are all tied to different parts of the same ecosystem, the total cost can rise quickly and become harder to predict.

• Can your architecture evolve? When business logic, security models, and semantic context are locked into one environment, moving workloads later is not a simple switch. It can mean rebuilding the very foundations your analytics and AI depend on and limiting ways to interact with vendors you selected to give you an advantage.

• Who controls the pace of your AI strategy? If AI agents can only work with your enterprise data through vendor-approved routes, then your AI program moves at the speed, scope, and direction that vendor allows.

Freedom Is Not a Feature. It's an Architecture Decision.

At Qlik Connect® a few weeks ago, we talked about three things enterprises need to make AI work in the real world: context, trust, and freedom. The first two get a lot of airtime. The third one is where the real battle is being fought, and it's the one SAP's Sapphire agenda has no interest in discussing.

Freedom means the ability to adapt as technologies, regulations, and business requirements change. Qlik is focused on increasing optionality for customers building for AI, giving them the freedom to use their data across the platforms, clouds, tools, models, and agent ecosystems they choose. SAP’s current strategy appears to move in the opposite direction, decreasing optionality by concentrating more control inside SAP-defined pathways. In practice, that means your data architecture cannot be owned by any single vendor’s commercial roadmap.

You don't want your data to be a hostage. You don't want to be making decisions in your architecture that aren't good for the business, choosing a data platform or an AI solution because it's the easiest path, or because a vendor already has you and makes it difficult to leave.

That is exactly the architecture SAP has spent the last several weeks assembling. And at Sapphire, they will call it openness.

Freedom, concretely, means your data lands in infrastructure you fully control. It means open formats, Iceberg, Delta Lake, that let you use the best AI and analytics tools available today without waiting for any vendor to certify them. It means decoupled storage and compute, so you can bring the right engine to the right workload rather than the only engine your vendor permits. It means being the trusted data and reasoning substrate that sits beneath whichever agent ecosystem you choose, not the one your ERP vendor has approved.

Freedom is not anti-SAP. It is pro-enterprise. The organizations that will move fastest on AI are the ones whose data is portable, governed by them, and genuinely free of vendor restrictions on where it can go and how it can be used. Not free in the marketing copy. Free in the architecture. Free in the contract. Free to leave.

“Open” for business? Think again.

SAP will use Sapphire to tell you they are open. They published a policy saying otherwise three weeks ago. SAP will tell you they are building for the customer. They bought the platforms customers were using to stay independent. SAP will tell you their architecture enables AI freedom. They wrote a clause prohibiting your AI agents from accessing your own data without their permission.

Words are easy. Policies, acquisitions, and API restrictions are decisions. And decisions don't stop when the conference ends.

Don't judge SAP by what you hear in Orlando. Judge them by what they do before they get there, and what they do after they leave.

If you want to see what enterprise AI looks like when the data underneath it is genuinely open, truly yours, and built for freedom, we made that case at Qlik Connect. The keynotes, the customer stories, and the proof points are available on demand.