Enterprise software decisions are not like other technology decisions. Timescales are longer, dependencies run deeper, and the cost of reversing course – if reversing course is even possible – tends to be measured in years and nine-figure sums. Which is why the relationship between a major ERP vendor and its customers is, more than almost any other in enterprise technology, built on trust. Not just trust in the product, but trust in the roadmap, the pricing model, and the terms under which you’ve committed.
For SAP customers, that trust has been tested repeatedly over the past decade. Here’s the familiar pattern: a new platform is announced, the business case is made, and customers commit at scale. Then, once dependency is established, the terms change. Not always dramatically, not all at once – but change they do, and in a direction that lines up with SAP’s needs rather than those of the customers who signed the contracts.
So, let’s set that record plainly, because right now SAP is asking its customers to trust it again. The past decade has given us enough evidence to ask whether that trust is well placed.
If you were an early mover, you would have signed up for S/4HANA, SAP’s next-generation ERP suite, when it launched in February 2015. ECC 6 customers were encouraged onto the new platform with the promise of innovation, performance, and a long runway behind both. SAP had committed to maintaining and innovating on S/4HANA until 2040.
The business case was the kind that took months to build and required board-level sign-off, with TCO models and innovation roadmaps attached. That twenty-five-year horizon was the foundation for many of those decisions.
Then, in July 2023, SAP announced it would no longer be innovating on the on-premise version. The future of S/4HANA was in the cloud, and specifically in RISE with SAP. Less than nine years into the twenty-five-year commitment, the roadmap customers had signed off on ended seventeen years early. The platform they had committed to became, almost overnight, a migration project.
It’s also worth understanding what S/4HANA on-premise actually was, because this matters to the scale of what customers were asked to accept. It also underpins some of the issues customers are now seeing in its cloud incarnation, RISE.
SAP took an existing product, mandated a move to its proprietary HANA database – replacing the Oracle, DB2, and SQL Server environments many customers had run for years, and positioned it as a new product requiring new contracts and new investment. ECC customers who had database choice built into their existing agreements found that choice removed entirely if they wanted the new S/4HANA suite.
S4HANA was not just an upgrade. It was a re-platforming that used HANA’s in-memory capabilities to eliminate redundant data structures, enable real-time insight, and modernise the user experience. The trade-off involved significant migration complexity, risk and expense for existing ECC customers who, more often than not, could not justify the move from what was, and still is, an exceptional product.
The product was new in name, but in substance it was an evolution that came with a transformation-sized price tag.
For the customers who did justify the move, the buy-in was to the promise of future innovation and a roadmap to business value that only S/4HANA was supposed to deliver. That roadmap was all but abandoned after nine years, leaving those customers with a choice: transform again or lose the roadmap they had bought into only a few years earlier.
A pattern, not a one-off
If you came in slightly later, the story is different in its specifics but identical in its shape. From early 2021, you moved to RISE with SAP, the cloud subscription programme through which S/4HANA is delivered as a managed cloud service, because RISE was presented as the future platform, the one that would be everything to everybody, fully cloud-native and AI-ready.
You committed on that basis. Then, by 2024, SAP had RISE Premium as the tier required to access AI capability. The platform you had signed up for had developed a new tollbooth between where you were and where you needed to be, and the only way through it was to spend more.
In April 2026, SAP went further. With no announcement and no customer briefing, it updated its AP Terms of Use to prohibit the use of SAP APIs to connect autonomous or generative AI systems to your SAP environment, unless those systems run through SAP-endorsed pathways. Your data, your system, your infrastructure – but the AI tools that can access it have to be those SAP has approved, with Joule chief among them. The pattern that took away your database choice on the move to S/4HANA has now taken away your AI agent choice.
This week, the position has changed again. Joule assistants and agents would, after all, be made available to on-premise S/4HANA and ECC customers, which on the surface looked like a softening of the cloud-only innovation stance taken in 2023. But the conditions told a different story. To access AI from on-premise, customers would need to commit at least 50% of their maintenance spend to the cloud and sign up to a new commercial tier, the Max Success Plan.
In other words, on-premise customers could now have AI, but only if they agreed to spend more with SAP and sign a new contract on SAP’s terms.
Every two to three years, commercial terms move in a direction that suits SAP. And those moves tend to come at the point when customers become too dependent on the platform to push back effectively. It has happened repeatedly, across different products and different customer groups, and each time the result has been the same: less customer choice, more vendor control.
First, it was database choice, then the innovation roadmap. Now it is AI agent choice. Each time, the right to flex, to swap, or to keep your options open is what gets removed.
The question I would encourage any customer currently evaluating its next SAP commitment to ask is not whether SAP will move the goalposts again, but where, and what part of their stack will be the next to stop being theirs to choose.
Why AI is the next lever
The next move is being made through AI. And to understand why, it helps to understand the position SAP is in.
RISE migration has been slower than SAP needs. What customers have been asked to undertake – rearchitecting core systems, retraining teams, rebuilding processes around a cloud model – comes with a price tag that, for large enterprises, runs into hundreds of millions. Customers who have built a financial case for it have largely come back with the same conclusion: the return is not there.
SAP’s response has been twofold. First, it sets a deadline. Mainstream support for ECC ends in 2027, after which customers remaining on the platform face escalating costs and diminishing coverage. Then it introduced Joule, SAP’s AI assistant, positioned as the capability that finally makes the case for moving. AI is the new reason, and RISE is the way to access it.
But before committing to that roadmap, there are questions worth asking carefully.
The first concerns data. If you run your operations through SAP’s AI tooling, the data those tools process is not, in any straightforward contractual sense, fully yours. The details are buried in end-user agreements, and the specifics will vary by contract. But it is worth asking your legal team directly whether your data could be used to train and refine SAP’s AI capability and ultimately made available across its wider customer base, including your competitors.
The second is a technical problem that SAP has not been resolved. Central to the RISE proposition is the concept of clean core, the principle that customers standardise their SAP environment by removing the customisations that have built up over years so they can run S/4HANA on a clean, structured data foundation. Joule is built to operate on exactly that.
But the reality in almost every large SAP environment is the opposite. Decades of customisation, bolt-ons, and workarounds mean the underlying data estate looks nothing like the environment Joule is designed for. Customers are being sold an AI capability calibrated to a version of their business that doesn’t exist, and they are being asked to spend heavily to reach it.
And then there is what it will cost when they do.
SAP has confirmed it is moving away from subscription-based licensing towards consumption-based pricing, driven by the reality that AI agents will increasingly replace the human users on which per-seat models have historically been built. In practice, Joule is already metered. Every interaction consumes SAP AI units. The base allotment included in RISE contracts is typically insufficient for production-scale deployment, and overage charges run at significantly higher rates than contracted pricing.
None of those features in the business cases being signed today. It is a pricing model whose full implications will only become clear once customers are already too committed to renegotiate, which, if the pattern holds, is exactly the point.
The choice most customers don’t know they have
Here is the question I put to customers weighing up their next SAP commitment: if you took the programme cost of a RISE migration and, instead of spending it on an ERP upgrade, invested it in building the technology capability your business actually needs, on platforms and architectures you control, what would that do for your business?
Staying on a well-supported, stable ECC environment frees up both capital and internal capacity for exactly that kind of decision-making. Migration is not, regardless of how SAP’s sales cycle frames it, the only option available.
I’m not arguing that every SAP customer should stay put. For some organisations with simpler environments and genuine strategic alignment with the RISE model, it may be the right call. But that decision should come from a clear-eyed assessment of business need and an honest accounting of what the commercial relationship with SAP has historically looked like, not from a deadline that exists because the business case couldn’t make itself.
SAP builds software that runs some of the most complex organisations in the world, and ECC remains in production across the majority of large enterprises for a very good reason: it is good software. That has never been in doubt.
But what SAP is offering customers right now is theory, concept, and promise. The past decade offers a factual record. At some point, those two things have to be weighed against each other honestly.
So, are you going to base your business’s future on theory or on fact?
