11 Aug 2026 by Thomas Evans
If you are still running SAP on‑premises in 2026, it may feel like you are saving your organisation money by sweating those enterprise resource planning assets for as long as possible.
But as a services partner who lives in both worlds – helping businesses manage and maintain on‑premise ECC and S/4HANA Cloud platforms – I’m convinced the inflection point has arrived. Staying on‑prem now carries more risk, more cost and more strategic downside than making the move to SAP’s cloud platform.
On‑prem SAP customers like to think of themselves as “stable”. In reality, they are increasingly frozen in time. They are actively opting out of a lot of new capability, stronger security and the AI-powered innovation that competitors are already exploiting.
In the SAP world, even with the best project team and a fair wind behind you, most organisations manage a major upgrade every two years at best. By the time you’ve justified the business case, procured hardware, scheduled downtime and fought your way through testing, the “new” release you’ve landed on is already six months or more out of date.
Contrast that with SAP S/4HANA Cloud, public edition, where innovations – including AI-driven ones – are delivered twice a year as part of your service. Those releases bring new functionality, usability improvements and entirely new solutions, such as new commerce capabilities that plug straight into your core ERP without the “big project” overhead an on‑prem customer would face.
Cloud is where SAP’s AI actually lives
A big part of the “why now?” in 2026 is simple: SAP’s most powerful AI capabilities are being built for, and delivered through, its cloud platform.
SAP is embedding AI across S/4HANA, HCM, CX and other cloud solutions – everything from embedded copilots like Joule, to document understanding, to predictive analytics grounded in your transactional data. Base AI capabilities are included with cloud subscriptions, so day‑to‑day intelligence comes as part of the package, while more advanced SAP Business AI can be turned on, delivering measurable value.
On‑prem, you simply do not see that pace of AI innovation. You might integrate a point solution here or a chatbot there, but you miss out on the tight integration, consistent user experience and shared security and data governance model that comes from AI being part of the platform itself.
Security: DIY vs industrial‑grade
Perhaps the biggest reason the equation has flipped in 2026 is security.
If you are running SAP on‑prem, your security posture is constrained by two things: time and budget. You can buy a dozen tools, but somebody has to implement, patch and monitor them, and that “somebody” is competing with every other priority your IT team has.
Meanwhile, attackers – increasingly equipped with powerful AI tools – are probing your environment for the misconfigured system, the unpatched interface, the forgotten test box sitting on an exposed network.
We are on the cusp of legislation in New Zealand that could make boards criminally and financially liable for major data breaches, following the Australian model with potential multi‑million‑dollar fines. For many organisations, the prudent, conservative choice is now to move to a platform where security, patching and AI governance are handled by a vendor whose entire reputation depends on getting those things right.
SAP has to get security right. It has too many customers counting on it to secure their mission-critical data. That means layered controls – encryption, identity and access management, continuous monitoring, compliance attestations – all documented in the SAP Trust Centre for customers and auditors to inspect. You are effectively buying into an industrial‑grade security stack that would be very difficult, and very expensive, to replicate on your own.
The race to embrace AI makes this even more critical. In on‑prem environments, we are seeing enthusiastic teams spinning up their own models, and exporting data into uncontrolled tools, potentially punching holes in the organisation’s security defences in the process. In the SAP cloud, AI lives inside the same security and compliance framework as your core ERP, with access controlled, logged and auditable.
In other words, you bring AI back under central control, without stifling innovation.
Avoiding AI “bill shock” with visibility and control
We’ve all heard the horror stories about token usage for AI services “getting out of hand” on generic hyperscaler platforms.
SAP’s model is different, and as a customer or partner you have tools to keep it under control.
First, at the core application level, licensing for S/4HANA Cloud, public edition is structured around users and roles – advanced users vs self‑service, for example – which you can see and manage in detail via the SAP for Me portal. You can see, month by month, who is using what, which roles have been over‑provisioned and where you are underutilising licenses you are already paying for.
Second, for premium AI capabilities, SAP uses “AI Units” as a form of virtual currency. You buy a pool of AI Units annually, can model your expected usage with online calculators and then track consumption in SAP for Me so you can see, in real time, how many units you have used, where and on what. Overages are clearly defined and priced, which means “bill shock” is only likely if you are not looking at the dashboards SAP already provides.
Most importantly, you can act before costs get out of hand. If you see, for example, that a particular AI‑driven process or chatbot is consuming more than expected, you can adjust roles, tweak usage patterns or revisit the business case to ensure the value justifies the spend. That is a very different world from the opaque consumption models many organisations have faced with generic AI APIs.
What about data sovereignty?
Data sovereignty and residency concerns are real, particularly for customers in the public sector, and those managing sensitive health and financial data.
That’s why SAP has made significant investments in making Sovereign Cloud capabilities available to government departments and large enterprises across Australia and New Zealand. SAP has worked with public agencies across the region for over 35 years as a trusted partner. Users of SAP S/4HANA Cloud, public edition have the ability to select a geo-location, which allows them to keep data and applications stored in the ANZ region, typically in Australian data centres.
The SAP Sovereign Cloud On-Site solution which became available across ANZ in September, offers increased sovereignty for public sector customers and businesses they work with.
2026: the tipping point
From where I sit, working with customers who are already seeing the benefits, the bigger risk now is not moving too fast to the SAP Cloud – it is moving too slowly.
We’ve seen what continuous access to new functionality, and embedded AI governed inside a strong security framework can achieve for New Zealand organisations.
With mainstream support for SAP ECC scheduled to end in 2027, most customers are now on their migration path to the cloud.
If you want to accelerate that journey and lay the groundwork for a successful migration, get in touch. With hundreds of successful cloud projects beyond us, Realtech is here to help.