S01.1 · Technology & Digital Infrastructure
A $900B+ horizontal SaaS market where agentic AI is starting to erode the seat-based pricing model that funds it.
Enterprise Software & Applications is the horizontal SaaS market — ERP, CRM, HR and the other cross-functional applications sold to every industry at once. Gartner sized it at $900B in 2024, heading toward roughly $1.0T in 2025. The number that matters, though, is not the top line. It is the seat count underneath it. Two decades of per-seat pricing have rested on the assumption that more work means more licensed humans, and agentic AI is now automating the support and implementation labor that assumption was built on.
Growth has run near 11% (Gartner, 2024) and should hold at 10-12%, but its composition has changed: the driver is AI feature attach and price uplift on the installed base, not new logos. That is expansion within accounts — an NRR story, not a bookings story — and it makes retention metrics, rather than pipeline, the place to judge vendor health. Revenue concentrates in North America (~45%) and Europe (~25%); engineering sits in the US, India, Israel and Eastern Europe.
Microsoft, SAP, Oracle, Salesforce and Adobe together hold a combined double-digit share, with thousands of sub-scale mid-market ISVs beneath them. The long tail is the segment's defining structural fact. At 70-85% gross margins, over 80% recurring revenue and minimal regulatory gating beyond data privacy, a sub-scale ISV is cheap to run and cheap to integrate, which is why the tail has been the segment's consolidation story for years. Upstream, the segment rests on cloud infrastructure and developer tooling; downstream it feeds systems integrators, MSPs, and the embedded-finance layers that extend core applications into transactional revenue.
The two visible adjacency moves are worth separating. Pushing into vertical software opens spend the horizontal base never addressed — genuinely new dollars. Bundling security into the core platform opens nothing; it exists to block point-solution attach and hold the renewal, and it is deflationary to every standalone vendor whose feature it absorbs.
Agentic AI cuts deeper than either move. First, at the cost line: tier-1 support and implementation consulting are being automated, which hits the services and channel economics around these platforms before it touches license revenue at all. Second, at the pricing unit itself. When agents do work that named users used to do, the billable seat disconnects from the work performed, and a vendor that cannot re-price toward consumption or outcomes is left charging for a base that shrinks even as task volume grows. The task volume is not the problem. The unit of charge is.
The boundary already moving is orchestration. Buyers have started sourcing the layer that coordinates, governs and audits agents across systems separately from the applications it sits on — a budget line forming now, in live RFPs where generative-AI parity is already table stakes, not a forecast item.