SAP CFO Warning: AI Returns Require Moving Beyond Chatbot 'Low-Hanging Fruit'

SAP Finance Chief Dominik Asam declares that the era of relying on chatbots and coding assistants as the primary drivers of artificial intelligence value is ending, arguing that substantial returns require embedding AI into complex business processes where clean data and cost control take precedence over raw model power.

The Shift from Hype to Hard Value

Following the release of second-quarter results, Dominik Asam highlighted that while companies have poured capital into generative AI, evidence of broad productivity gains remains elusive. SAP argues that the real ROI lies not in general-purpose models but in governed systems embedded in specific operations.

■The "lion's share" of current AI token consumption is spent on "low-hanging fruits" like coding assistants, where the risk of error is minimal.

■Applying AI to finance or supply chains is exponentially harder because errors compound statistically over multiple steps, requiring "excruciating assurance levels" to meet compliance standards.

The Economics of Data Governance

Asem emphasized that the belief AI can magically solve issues within messy, legacy data silos is a misconception that leads to "extremely high token costs." The future of enterprise AI, according to SAP, lies in specificity rather than brute force.

■Companies must govern their own data to allow AI to operate with institutional knowledge, rather than relying on generic plug-and-play models.

■In practice, enterprises will utilize the "cheapest reliable tool"—whether open-source or a frontier model—that can deliver the required outcome safely, rather than defaulting to the most expensive option available.

From a macro strategy standpoint, this commentary serves as a reality check for the AI trade. We are seeing a shift from 'growth at all costs' to 'ROI-focused deployment' in the enterprise software space. If the integration of AI into legacy systems proves more capital-intensive than anticipated, we could see a contraction in the earnings multiples of AI-exposed tech stocks. The market is no longer paying for potential; it is paying for execution and margin expansion. Asam's emphasis on cost control suggests that the 'AI capex supercycle' might face efficiency headwinds in the near term.