SAP and Google Cloud have strengthened their partnership at SAP Sapphire 2026 with a series of innovations aimed at taking enterprise Artificial Intelligence beyond generic assistants. The declared goal is to connect critical business data with acting agents that can operate on real processes, without forcing companies to continually move, copy, or rebuild the information already stored in their SAP systems.
The underlying idea is simple to explain but challenging to implement: many large enterprises have valuable data trapped in silos, outdated integrations, and manual extraction processes. This restricts the usefulness of Artificial Intelligence because an agent can only make good decisions if it understands the business context, relationships between entities, internal rules, and the actual state of operations. SAP and Google Cloud aim to bridge this gap with a more unified database and connected workflows linked to Gemini, BigQuery, and the new SAP Business AI Platform.
Agents that communicate between SAP and Google Cloud
One of the most significant announcements is the open collaboration between agents. SAP will integrate new agent capabilities into the SAP Business AI Platform to enable bidirectional communication between Joule and intelligent agents built on Google Cloud, such as the Gemini Enterprise Agent Platform and Gemini. This signifies a major shift: companies don’t want isolated agents by provider, but systems that coordinate across platforms without losing governance or traceability.
Interoperability will be crucial. An SAP agent might understand a purchase order, a financial issue, or a supply flow. A Google Cloud agent can query data, generate analyses, trigger processes, or work with external information. If both environments communicate securely and share context, the result can be more valuable than a simple conversational layer over the ERP.
SAP has already introduced at Sapphire its concept of “autonomous enterprise”, supported by SAP Business AI Platform, SAP Autonomous Suite, and Joule. The company aims for its agents not only to respond to queries but also to execute end-to-end processes across finance, supply chain, procurement, HR, and customer experience. Google Cloud joins this strategy as a partner for data, analytics, agents, security, and computing capacity.
The risk, as with many such announcements, lies in confusing vision with actual deployment. Existing technical components or preview features do not mean a company can automate critical processes without prior work. Data quality, permissions, workflow redesign, compliance controls, legacy system integration, and clear governance over each agent’s scope are essential.
BigQuery and SAP Business Data Cloud, without copying data
The most tangible piece of the announcement is SAP BDC Connect for BigQuery, currently in private preview. This integration allows semantically enriched SAP data to be shared directly with BigQuery in a bidirectional, zero-copy manner, without incurring costs for data duplication, according to Google Cloud. Practically, it aims to avoid one of the most common burdens in enterprise analytics projects: moving large volumes of data between systems, duplicating, cleaning, and recontextualizing it.
For organizations already using SAP as their transactional core and Google Cloud for data platforms or AI, this proposition is compelling. BigQuery can operate on enterprise-meaningful data while SAP retains control of context and governance. This can reduce manual pipelines, lower technical debt, and accelerate analytics, predictive modeling, or automation projects.
Google also highlights Cortex Framework, now in preview, as a set of accelerators to transform fragmented data into richer, context-aware data products. The goal is to turn technical database structures into a semantic layer that is more understandable for users, analysts, and agents. In other words: moving from difficult-to-interpret tables to business concepts that AI can use with less risk of error.
This point is critical to reducing hallucinations. A model working with business data devoid of context can produce plausible but incorrect answers. An agent connected to a governed database, with semantic meaning and clear rules, is more likely to operate accurately. While it cannot eliminate all risks, it can significantly mitigate them.
More memory for SAP HANA and a focus on sovereignty
The alliance also includes infrastructure enhancements. Google Cloud announced the X5 series, with instances optimized for up to 48 TB of memory—50% more than the previous limit of 32 TB in the X4 family. This allows major SAP HANA and RISE with SAP customers to scale critical databases on a single node without redesigning applications.
Google cites Mercado Libre as a case study. The Latin American company states that the new 48 TB instances enable critical databases to grow on a single node, allowing it to avoid major redesigns as it advances in leveraging BigQuery, Gemini, and RISE. This example illustrates that enterprise AI depends not just on models but also on ample memory, performance, and data capacity for real-time operations.
In cloud sovereignty, SAP and Google Cloud announced a collaboration with S3NS to deploy RISE private cloud on a SecNumCloud-certified platform in France. This is especially relevant for regulated organizations like Thales, which need to transform their ERP systems without exposing certain data or workloads to environments under strict national controls.
Digital sovereignty is gaining importance in Europe. It’s not enough to just use cloud services; many organizations want clarity on where their data is processed, applicable legislation, access controls, and service certifications. In this context, the combination of RISE, S3NS, and SecNumCloud aligns with a clear trend: enabling cloud modernization in sensitive sectors without adopting overly generic control models.
Specific security for SAP environments
Another update is Google SecOps for SAP, available in preview. Google and SAP position it as a security operations layer with AI-driven agents and threat detection tailored to SAP applications, integrated into the broader IT environment.
Security in SAP systems is critical because many of these systems underpin finance, procurement, logistics, payroll, production, and supplier relations. An incident within SAP can impact billing, inventory, compliance, or operational continuity. Integrating SAP signals into a broader Security Operations Center can help detect patterns that were previously scattered across application, security, and infrastructure teams.
The agent-based approach should be handled with caution. Automating detection, investigation, and response can improve response times but requires controls. An improperly configured agent may generate unnecessary alerts, prioritize incorrectly, or take overly broad actions. Credibility depends on rules, traceability, human oversight, and clear distinctions between recommendations and automatic execution.
SAP and Google Cloud are working to overcome one of the main barriers to enterprise AI: transforming business data into reliable actions. The innovation is not just in Gemini, Joule, or BigQuery individually but in connecting these with governed data, sufficient memory, targeted security, and a collaborative architecture for agents.
Autonomous enterprise will not be achieved just by toggling a feature in an admin panel. It will come from having trustworthy data, well-defined processes, interoperable platforms, and clear boundaries on AI decision-making. The SAP-Google Cloud partnership is a significant step toward that goal, but its true impact will depend on how companies translate these announcements into measurable, secure projects.
Frequently Asked Questions
What have SAP and Google Cloud announced at SAP Sapphire 2026?
They have reinforced their partnership with integrations between Joule and Gemini agents, SAP BDC Connect for BigQuery, new 48 TB instances for SAP HANA, Google SecOps for SAP, Cortex Framework, and sovereign cloud options with S3NS in France.
What is SAP BDC Connect for BigQuery?
It is a private preview integration that enables sharing semantically enriched SAP Business Data Cloud data with BigQuery via bidirectional, zero-copy access, avoiding massive data duplication.
Why are the 48 TB instances important?
They allow running large SAP HANA databases on a single node, which is crucial for RISE with SAP customers with critical operations and rapid data growth.
What role does Gemini play in this alliance?
Gemini and Gemini Enterprise Agent Platform will serve as part of Google Cloud’s agent ecosystem, enabling collaboration with Joule and usage of governed business data in workflows.
via: cloud.google