SAP's Q2 2026 results told a familiar story about cloud growth but more revealing was what CEO Christian Klein chose to talk about on the company earnings call. Rather than dwelling on the numbers, he argued that the next phase of enterprise AI will depend less on model capability and more on access to trusted business data and processes.
The headline figures were solid. Cloud revenue was up 24% at constant currencies to €6.3 billion, current cloud backlog growth has accelerated to 26%, and total revenue is up 11% to €9.9 billion. Non-IFRS earnings per share missed forecasts, however, hit by higher R&D spending and AI token costs as SAP scales its new Business AI platform.
Christian Klein, CEO of SAP, outlined the main drivers of these results: "Customers are choosing SAP to enable accurate and compliant AI outcomes grounded in their most critical business processes and data". He also explained that AI and SAP Business Data Cloud were "embedded as key pillars in more than 90% of our 50 largest deals". Those comments build on the strategy SAP first outlined at Sapphire 2026, where it introduced its vision of the Autonomous Enterprise and a unified Business AI Platform designed to ground AI in enterprise data and governance.
Klein wasn't really talking about SAP's models. He seemed to be talking about trust and the idea that AI becomes genuinely useful only once it understands how a business actually works.
Enterprise AI is becoming a battle for business context
Foundation models are rapidly improving, making them less of a differentiator on their own. As the leading models converge on similar levels of capability, enterprise software vendors are increasingly competing on something else. The question is who can give AI the deepest understanding of a customer's business. That doesn't mean models stop mattering but it does suggest that the more interesting competitive front is shifting towards the data, workflows and governance surrounding them. It's the same principle behind grounding AI responses in an organisation's own knowledge rather than relying solely on a model's general training data.
Picture the difference in practice. An AI agent that can see a customer's last three orders, their warranty status, payment history and open support tickets will generally outperform one relying purely on a general-purpose model with none of that context. If SAP's earnings are any indication, that's where enterprise AI competition is heading next.
Gartner sees the same trend. Earlier this month, in a press release on AI platform spending, analyst Arunasree Cheparthi said "enterprise AI budgets are coming under greater scrutiny, with increased focus on usage efficiency, cost control and measurable outcomes." The critical implication here is that enterprise AI is more and more being judged by the business impact it delivers.
Why this matters for customer experience
The question many CX and IT leaders are wrestling with isn't whether AI can hold a fluent conversation. It's whether it can safely resolve a billing dispute, check inventory or update a customer record. Those tasks depend on access to CRM data, order history, policies, billing systems and permissions, and on governance that defines what AI can access, what actions it can take and when it should escalate to a human. Without that context, AI agents can answer questions but stall the moment a customer needs something actually done. With it, they can resolve enquiries, personalise responses and stay within compliance boundaries while doing so.
Many brands have concentrated agentic AI on low-risk tasks such as feedback collection and appointment reminders. Yet the harder journeys, including returns and refunds, remain the least automated, largely because fragmented systems leave AI without the context to finish what it starts.
A wider shift across enterprise software
SAP isn't alone. Microsoft, Oracle, ServiceNow and Zoom, the last through its recent acquisition of Common Room, have all been building out data and context layers around their AI agents over the past year. Salesforce has been making the same case publicly, recently connecting fan, host city and stakeholder data across its Agentforce 360 portfolio during its FIFA World Cup partnership. SAP's own acquisitions of Dremio, Reltio and Prior Labs point the same way, towards stronger data foundations, cleaner master data and more accurate predictions for the agents built on top of them.
These moves suggest an enterprise AI market increasingly converging on the same assumption, that agents are only as useful as the business context they can draw on. Five years ago, enterprise AI was largely a race to access the best model. Now it seems the next race will be over something far less visible: who owns the most useful business context.