Daikin Europe is moving its SAP environment onto RISE with SAP and Amazon Web Services as part of a four-year extension of its relationship with Kyndryl, putting a large enterprise-systems migration beneath plans for more data-driven operations and artificial intelligence.

Kyndryl will lead the migration and continue managing infrastructure services around Daikin Europe’s SAP estate, which supports manufacturing, supply chain, warehouse operations, human resources, sales, and other business functions across Europe, the Middle East, and Africa. The companies have worked together since 2012, making the programme a long-running operational modernisation rather than a clean-sheet cloud deployment.

Under the new agreement, Daikin’s SAP ERP landscape will move to RISE with SAP using AWS as the underlying cloud platform, while Kyndryl will provide integration and managed services connecting SAP with the manufacturer’s wider technology estate. Kyndryl Bridge will also be used for monitoring, automation, and operational visibility.

The arrangement illustrates a less conspicuous part of enterprise AI adoption because companies can buy access to increasingly capable models quickly while the systems containing their operational data often take years to modernise. Manufacturers in particular depend on ERP applications spanning procurement, inventory, production, finance, logistics, and sales, leaving AI projects dependent on whether those underlying records can be accessed, reconciled, governed, and trusted.

ERP deadlines meet AI ambitions

Daikin’s programme also arrives as long-standing SAP customers face a narrowing maintenance timetable. SAP has committed mainstream maintenance for core Business Suite 7 applications until the end of 2027, with optional extended maintenance available until the end of 2030, giving organisations still running older ERP estates a practical reason to determine how and when they will move.

The migration question is wider than replacing one software version with another. ERP systems accumulate interfaces, custom code, business rules, data structures, and dependencies over many years, while manufacturing companies often connect them to planning systems, warehouses, production applications, supplier networks, and local operations spread across several countries.

Moving that environment can simplify parts of the estate, but it can also expose years of technical debt. Data that are adequate for a monthly financial process may not be consistent enough for automated decision-making, while interfaces built around older workflows become obstacles when companies attempt to introduce real-time analytics or agent-based systems.

AI consequently raises the stakes around work that might previously have been described mainly as cloud or ERP modernisation. A model asked to explain inventory risk, anticipate supply problems, or support a procurement decision needs access to current and correctly structured information, while an AI system allowed to take actions inside operational software introduces additional questions around permissions, auditability, and error handling.

Daikin’s decision to use Kyndryl as a central point of coordination between its own estate and SAP reflects another persistent feature of large transformation programmes. Organisations rarely operate a single cloud or enterprise-software environment, and the difficult work usually lies in connecting platforms that were purchased at different times for different purposes.

AI readiness starts below the model

Technology suppliers increasingly describe modernisation programmes as AI-ready, although the phrase can cover considerably different levels of capability. Moving applications onto newer infrastructure does not in itself produce useful AI, and neither does adding an AI-powered monitoring layer to an existing estate.

What the migration can provide is a cleaner technical foundation on which later AI projects are easier to build. Standardised interfaces can make operational data more accessible, cloud infrastructure can provide greater flexibility for analytics workloads, and better observability can reduce the risk of adding automation to systems whose behaviour is poorly understood.

That distinction is especially relevant in manufacturing, where adoption is constrained by operational continuity. An office productivity tool can often be introduced to a group of employees and withdrawn if the results disappoint; an ERP migration touching warehouses, production planning, and supply chains has to survive much more demanding change controls because disruption can stop physical goods moving.

The four-year term also cuts against the idea that enterprise AI adoption is principally a race to deploy the newest model. Daikin’s systems work will unfold over a period in which AI products, model providers, and software interfaces are likely to change repeatedly, while the ERP environment beneath them will be expected to remain operational throughout.

That leaves integration architecture and data governance carrying more strategic weight. Companies that tie an AI programme too tightly to one model generation may find their assumptions obsolete quickly, whereas investment in cleaner data, clearer interfaces, stronger identity controls, and modernised core applications can support several generations of tooling.

For Daikin, the immediate programme is still an SAP transformation rather than an AI deployment. That distinction makes it a useful enterprise case study: before AI can act on manufacturing, warehouse, supply-chain, or sales information, the systems containing that information have to expose it reliably without destabilising the operations they already run.