S01.4 · Technology & Digital Infrastructure
Data and analytics platforms alongside a much larger AI/ML spending pool, with agentic tools already displacing traditional BI.
Data, Analytics & AI/ML spans data-platform and business-intelligence software plus the far larger applied-AI spending pool beside it. Core data and analytics platforms are sized at $300-490B for 2025 (Grand View Research and Fortune Business Insights, differing on scope); Gartner forecasts $1.5T of total worldwide AI spending in 2025, $644B of it generative AI. The number to hold onto is none of these: it is that agentic tools are already displacing dashboard-based business intelligence in production — a substitution event inside the segment, not a rising tide under it.
Analytics has grown ~15-20% historically; AI-specific software is forecast at 25-35% on generative-AI adoption. Revenue concentrates in North America and Europe; the talent base remains centered in Silicon Valley and is spreading into China and India. The platform layer belongs to the hyperscalers plus Databricks, Snowflake and Palantir, with a long fragmented tail of point analytics and MLOps vendors underneath — a tail whose survival question is the standard one: does the buyer renew this separately in three years, or has a platform absorbed it?
Platform incumbents earn high margins, but capital intensity is rising with training-compute requirements, and pricing has gone hybrid — part subscription, part consumption — which shifts revenue risk toward usage in a way pure-subscription analytics never carried. Upstream, the segment depends on cloud infrastructure and GPU supply. Downstream it feeds vertical AI applications and decision-support users. Regulation is tightening: the EU AI Act and sectoral privacy rules now impose compliance costs this segment never used to bear.
The extension moves point in different directions. Vertical AI applications are new revenue built on data a vendor already holds. Governance and security bundling is a compliance moat around regulated data — worth having, but it defends existing revenue rather than adding any.
Agentic AI hits both a cost line and a product category. The cost line is data engineering and analyst report-building — headcount that has always sat around these platforms. The product category is the dashboard itself: natural-language query answers the question directly, bypassing the pre-built reports business-intelligence vendors have sold for two decades. A vendor whose revenue is the report, not the data underneath it, is on the wrong side of that substitution.
The boundary being redrawn splits the segment in two: data infrastructure — the platforms that store, govern and move data — and AI application — the tools that act on it. Different vendor bases, different competitive dynamics, and the split is running in production deployments now, not on a multi-year horizon.