Building a strong data infrastructure for AI agent success
AI Summary
According to McKinsey's annual AI report cited in a March 10, 2026 MIT Technology Review article, nearly two-thirds of companies were experimenting with AI agents in late 2025, and 88% were using AI in at least one business function — up from 78% in 2024 — yet only one in ten companies have successfully scaled their AI agents. SAP Data & Analytics President and Chief Product Officer Irfan Khan argues that the primary obstacle to scaling is not model capability but inadequate data infrastructure, specifically the lack of business context delivered alongside data. A report from consultancy Deloitte found that only 40% of companies believe their data management processes are ready for AI, down from 43% the prior year, while the Institute for Data and Enterprise AI (IDEA) found that two-thirds of business leaders do not fully trust their data. The article highlights that more than two-thirds of companies cite data silos as a top AI adoption challenge, with more than half of enterprises managing over 1,000 data sources. Khan contends that agentic AI will not replace SaaS applications but will instead function as an additional engagement layer on top of existing software stacks, requiring a semantic or 'business-fabric' layer to harmonize data across platforms such as Snowflake, Databricks, and Google BigQuery. He advises enterprises to prioritize data governance and semantic modeling before scaling AI pilots, and cautions against full automation of critical business processes too early given the oversight requirements involved.
Why it matters
The data points from McKinsey, Deloitte, and IDEA collectively signal that enterprise AI adoption is hitting a structural bottleneck in data infrastructure, which has direct implications for vendors in the data management, cloud storage, and AI platform spaces — including SAP, Snowflake, Databricks, and Google Cloud. The finding that AI readiness among enterprises is declining year-over-year, despite rising AI investment, suggests a growing market opportunity for data governance, semantic layer, and data integration solutions rather than for foundation model providers alone. Khan's assertion that SaaS applications will coexist with — rather than be displaced by — agentic AI is a notable counterpoint to a prevailing narrative among some investors and analysts that positions AI agents as an existential threat to the SaaS business model.
Scoring rationale
The article covers enterprise AI agent adoption and data infrastructure requirements with market relevance to platforms like Snowflake, Databricks, SAP, and Google BigQuery, but reads primarily as sponsored thought leadership rather than a direct market-moving AI story.
Impacted tickers
This summary was generated by AI from the original article published by MIT Technology Review AI. AIMarketWire does not provide trading advice. Always refer to the original source for complete reporting.