The forecast and what sits behind it
Gartner projects that 234 billion dollars of enterprise SaaS spending will be at risk by 2030 as AI agents take over work that people do through software interfaces today. The figure represents roughly 20 percent of enterprise SaaS spending by the end of the decade. The logic is direct: as agents interact with business systems through APIs while humans stop clicking through screens, the value of per-seat licensing and interface-centric design erodes. George Brocklehurst, a Managing Vice President at Gartner, captured the shift in a single line: "You are no longer buying software primarily for people; you are increasingly buying it for agents."
We take the forecast seriously because it names a mechanism, and not just a trend. Seat-based pricing assumes a human operator for each license. When an agent performs the same task programmatically, one integration can displace many seats, and the pricing model that funded the SaaS boom loses its anchor. Gartner is not predicting that enterprises spend less on software overall. It is predicting that a large slice of current spending shifts to different vendors, different pricing structures, and in some cases different categories entirely. For a CIO managing a large application portfolio, that reallocation is the strategic event, and it is close enough to plan for now.
Why seat-based pricing is exposed
The exposure concentrates in software priced by named user. Consider the categories where most of the work is repetitive interaction with a system of record: service desks, expense processing, basic customer support, and routine data entry across CRM and ERP. These are the functions agents automate first, and they are also where seat counts are highest. As an agent absorbs the interaction, the enterprise no longer needs a license for every person who used to do that work. The vendor keeps the underlying system, and the seat revenue that scaled with headcount comes under direct pressure from the same automation the vendor is promoting.
This creates an uncomfortable dynamic for incumbent SaaS providers. Their own agentic features can cannibalize their seat revenue, which is why several are moving deliberately toward consumption and outcome-based models. For the buyer, the practical implication is that renewal economics are about to become negotiable in ways they have not been for a decade. A CIO who understands which of their applications are most exposed to agent-driven displacement holds real leverage in the next contract cycle, and can push for pricing that does not penalize the automation the organization is pursuing across its estate.
The procurement questions this changes
Gartner's guidance to buyers reorders the criteria for evaluating software. The first shift is to weigh a product's API capabilities as heavily as its user experience, because an agent's ability to operate the system depends on programmatic access, and not on interface polish. The second is to scrutinize vendor contracts for restrictions on autonomous agent use. Some licensing terms written for human users explicitly or implicitly prohibit programmatic operation at scale, and an enterprise that deploys agents against such a system may find itself out of compliance. These are questions most procurement teams have never had to ask.
We would add these tests to every software evaluation starting now. Ask whether the vendor's API exposes the full functionality an agent would need, or only a subset that keeps humans in the loop by design. Ask whether the contract permits an agent to perform licensed actions, and at what volume. Ask how the vendor prices agent-driven usage as seats decline. Vendors that cannot answer clearly are telling you their model is not ready for the way you intend to operate. The enterprises that build these questions into sourcing today will avoid discovering the answers during an audit or a failed deployment later.
Data ownership becomes the leverage point
The forecast surfaces a subtler point about where durable value accrues. Gartner urges enterprises to prioritize ownership of their operational learning data, using a measure it calls the Knowledge Retention Rate. As agents run processes over time, they generate a record of how work actually gets done, which decisions were made, and what outcomes followed. That accumulated operational knowledge is a strategic asset. If it lives inside a vendor's platform and cannot be extracted, the enterprise has traded short-term convenience for long-term lock-in that deepens every quarter the agents keep running.
We think data ownership is the leverage point most enterprises will underweight until it is too late. The organizations that retain their operational learning can move between platforms, train their own models, and negotiate from strength. Those that let it pool inside a single vendor's system will find switching costs rising as the agents become more capable and more entangled with the vendor's tooling. This is a direct argument for the kind of captive capability and data control that enterprises running their own centers are positioned to capture. The decision about where operational knowledge lives deserves a deliberate choice made early.
How vendors are likely to respond
Vendors are not standing still, and their responses will shape the market a CIO negotiates in. Gartner's analysis points to a shift from seat-based toward outcome-based and consumption pricing, and to incumbents embedding agentic capabilities directly into their processes so that the value moves with the automation. The providers with the deepest data and the widest install base have the most to protect and the most raw material to build with. We expect the largest platform vendors to defend their position aggressively by making their own agents the easiest ones to run against their systems.
The opening this creates for challengers is real. Gartner notes that AI-native startups may come to dominate the orchestration layer that sits above individual applications and coordinates agents across them. That orchestration layer is where a great deal of future value will concentrate, because it is closest to the actual business process. For a technology leader, the strategic question is whether to let an incumbent own that layer as an extension of its existing footprint, or to keep it independent so that no single vendor controls how agents move across the estate. That architectural choice will outlast any individual contract.
What CIOs should do now
The concrete first step is to map your application portfolio against agent-driven displacement. Identify which systems carry the highest seat counts against the most repetitive interaction work, because those are the contracts where both the risk and the negotiating opportunity are largest. Overlay each vendor's agentic roadmap and pricing direction, and flag the contracts renewing in the next twelve to eighteen months. This exercise converts an abstract 2030 forecast into a specific, near-term action list, and it gives procurement the evidence to renegotiate terms before the shift is priced against you.
The second step is to set policy on API access, contract terms, and data ownership before individual teams cut their own deals. An enterprise that lets each function adopt agents independently will accumulate inconsistent licensing exposure and scattered operational data that is hard to consolidate later. We would establish a standard that every new software agreement addresses agent authorization, programmatic access, and data extractability. Gartner's 234 billion dollar figure is a forecast about the whole market, and the way it lands inside any single enterprise depends on decisions a CIO can start making this quarter. The organizations that treat it as a planning input will hold the leverage.
)
