SAP’s AI Vision: Keep The System Of Record In House, Not Outsource The Brain

SAP has introduced Joule, a new AI interface embedded across its core enterprise solutions, including S/4HANA Cloud and SuccessFactors. This move underscores SAP’s strategy to retain control over the data that powers AI, rather than relying on external models. The company’s focus on owning its data infrastructure aims to create a more secure, compliant, and contextually rich AI environment, marking a significant shift in enterprise AI development.

As of mid-2026, SAP reports that Joule is operational across more than 35 solutions, with over 30 specialized agents and 2,500 skills, and plans to expand to 50 agents and 200 skills by Q3 2026. The platform is supported by a €100 million partner fund aimed at enabling system integrators to develop custom agents using Joule Studio, SAP’s low-code-to-pro-code agent builder. These agents have demonstrated measurable outcomes, such as reducing HR process cycle times by 40-60%, cutting operational costs by 16%, and improving developer productivity by approximately 20%, according to SAP’s published data.

SAP emphasizes that Joule does not pull answers from open internet sources but instead reads structured, permissioned enterprise data from its Business Technology Platform. This approach leverages a Knowledge Graph to understand business-specific workflows and legal implications, creating a moat against competitors relying solely on large language models (LLMs). The architecture is model-agnostic, consuming third-party foundation models and orchestrating them within a unified layer, which SAP claims provides a competitive advantage by owning the data substrate rather than the models themselves.

Implications of SAP’s Data-Centric AI Approach

SAP’s strategy to keep the system of record in-house and focus on owning enterprise data could reshape how AI is integrated into large-scale business operations. By controlling the data layer, SAP aims to deliver more trustworthy, context-aware AI solutions that are less dependent on external models, reducing risks related to model quality, access, and costs. This approach positions SAP uniquely in the enterprise AI landscape, especially as hyperscalers and frontier labs concentrate on model development rather than data ownership.

For customers, this means potentially more secure, compliant, and reliable AI tools that integrate seamlessly with existing SAP systems. However, the strategy also introduces risks, including challenges in scaling adoption and managing variable AI costs, which could influence how quickly organizations operationalize Joule’s capabilities.

SAP’s Enterprise Data Dominance and AI Evolution

Most of the world’s business transactions—purchase orders, invoices, payroll, supply chain data—pass through SAP systems, giving the company a unique position in enterprise data management. SAP’s AI vision, articulated as “the Autonomous Enterprise,” hinges on leveraging this data to embed AI deeply into business processes. Prior to Joule, SAP’s AI efforts focused on integrating models into workflow automation, but the company now emphasizes owning and structuring its data environment to support AI’s contextual understanding.

The launch of Joule builds on SAP’s existing investments, including the acquisition of Prior Labs, and its development of a Knowledge Graph that maps business relationships. The company’s architecture aims to be model-agnostic, orchestrating third-party foundation models within its own data layer, thereby avoiding dependence on any single AI vendor or model quality. This strategic positioning contrasts with frontier labs’ emphasis on building larger, more capable models, highlighting SAP’s focus on data as a competitive moat.

“Our focus is on owning the data that AI models need, not just building the smartest models. That’s the core of SAP’s AI vision.”

— Thorsten Meyer, SAP AI strategist

Uncertainties Around Adoption and Cost Management

It remains unclear how quickly organizations will operationalize Joule’s capabilities at scale, given challenges in demand-side adoption. The €100 million partner fund indicates potential hurdles in incentivizing system integrators and customers to fully embrace the platform. Additionally, variable AI usage costs could complicate budgeting and forecast accuracy for CIOs and CFOs, potentially slowing adoption or leading to underutilization.

Further, the long-term impact of relying on third-party foundation models within SAP’s orchestrated layer is uncertain, especially if model quality or access conditions change unexpectedly.

Next Steps for SAP and Its Enterprise AI Ecosystem

SAP plans to expand Joule’s capabilities, with increased agent deployment and new integrations through its partner ecosystem. The company will likely monitor adoption rates closely, adjusting incentives and support structures to accelerate operational use. Additionally, SAP will continue refining its Knowledge Graph and model orchestration layers, aiming to strengthen its moat against competitors relying solely on open models. Monitoring how customers manage AI costs and integrate Joule into their workflows will be key in assessing the success of this strategy.

Key Questions

How does SAP’s Joule differ from other enterprise AI solutions?

Joule emphasizes owning and structuring enterprise data within SAP’s platform, using a Knowledge Graph for contextual understanding, and orchestrating third-party models—focusing on data control rather than building proprietary models.

What are the main risks associated with SAP’s AI approach?

Risks include variable AI costs impacting budgeting, slow adoption due to demand-side challenges, dependence on third-party models, and the complexity of integrating Joule into heavily customized existing systems.

Will SAP’s focus on data ownership limit its AI innovation?

While it may slow the pace of model development compared to frontier labs, SAP’s strategy aims to create more reliable, compliant, and contextually relevant AI solutions, which could prove more valuable in enterprise settings.

How might this strategy affect SAP’s customers?

Customers could benefit from more secure, integrated AI tools that leverage existing data, but they may also face challenges in managing variable costs and ensuring adoption across complex organizations.

Source: ThorstenMeyerAI.com