AI needs a strong data fabric to deliver business value

As AI shifts from experimentation to core enterprise workflows, a critical bottleneck has emerged. According to SAP's President and Chief Product Officer of Data & Analytics, Irfan Khan, the primary obstacle is no longer model performance or compute power, but the quality and business context of the underlying data. A recent survey indicates that by the end of 2025, half of all companies will deploy AI in at least three business functions, yet only 9% feel fully prepared with their data infrastructure. Khan warns that AI can generate answers quickly but, without proper context, it cannot exercise good judgment—a deficiency that directly impacts return on investment.

Traditional data strategies focused on aggregation into warehouses and lakes often strip away the crucial semantics that describe how a business actually works. This loss of context creates a critical shortfall for autonomous AI systems that must act on information, not just display it. For example, an AI managing supply-chain disruptions might have accurate inventory numbers but lack context on strategic customers or acceptable trade-offs during shortages, leading to technically correct but operationally flawed decisions. The emerging solution is a 'data fabric'—an abstraction layer that integrates data across applications and clouds while preserving business context, enabling AI to coordinate decisions safely and reflect real-world priorities.