AINEWS
Enterprise AI is shifting from pilot to fleet, and ServiceNow says the new bottleneck is cost control and oversight, with its AI Control Tower bidding to own the governance layer.
When cloud spend crossed from sandbox to line item, finance teams did not get a new platform. They got FinOps, the discipline of figuring out what the company was actually paying AWS, who was paying it, and which workloads could be turned off on a Friday. ServiceNow's chief financial officer told investors this week that enterprise AI is hitting the same wall, and the company is putting $1.5 billion of revenue target behind its bid to own the dashboard for it.
At the Citi TMT conference, ServiceNow said AI usage among ServiceNow's customers grew ninefold from the first quarter to the second. More than 50 of those customers now pay the company over $1 million each for new AI packages. The company raised its 2026 AI revenue target from $1 billion to $1.5 billion on the back of those numbers, and the gating problem, Mastantuono said, is no longer capability. It is governance: knowing which agents exist, what they cost, and whether they should still be running.
The product ServiceNow is selling into that gap is called the AI Control Tower. It is built to inventory the AI agents running across a customer's environment, including ServiceNow's own, third-party agentic platforms, and the ones internal teams have wired up themselves. It then does three things the rest of the enterprise AI stack largely does not. It tracks cost by department or by manager. It can turn agents off when they misbehave or stop paying back. And it reports the return on each AI initiative in a form a CFO can sign off on.
The numbers ServiceNow used to fill in the picture are vendor-reported. The company told the Citi TMT conference that the city of Raleigh cut IT service-desk costs by 65 percent after rolling out ServiceNow's agents, and that a European energy customer is tracking potential savings above $5 million. Neither figure has independent third-party verification, and ServiceNow did not provide one on the call. The pattern they describe, however, lines up with what Mastantuono said she is hearing from buyers. The conversation has moved from "how do we get more productivity out of AI" to "how do we run AI as a business line."
That shift is the underlying story, and it is bigger than any single vendor. When a new spend category crosses from pilot budget to platform budget, the same thing tends to happen. Cloud produced FinOps. SaaS produced license-management platforms and the chief SaaS officer. Mobile devices produced mobile-device management. Enterprise AI is now producing a similar layer, and ServiceNow is one of several vendors positioning for it. None of the others were on the Citi TMT stage this week.
The case for a real category is straightforward. The case for skepticism is sitting in ServiceNow's own stock chart. The same conference remarks that surfaced the 9x growth number came days after the so-called SaaSpocalypse, a sharp selloff in software stocks on fears that AI agents would erode SaaS pricing power. A governance product that tells an enterprise exactly which agents are worth the money is, by construction, also a product that helps an enterprise spend less on agents. The vendor is selling the metering on a category it has to convince investors still has room to grow.
There is a second tension the product does not resolve. A tool that can inventory every agent in an enterprise, turn them off, and audit their work is also a tool that decides what an enterprise is allowed to do with AI. That is governance in the FinOps sense, and it is also governance in the political sense. ServiceNow's pitch does not yet say who inside the customer gets to push the off button when the agent is doing exactly what it was told to do, and the answer will matter more than any of the numbers on the conference slide.
The next test is short. The 50 customers paying $1 million-plus have to renew and expand. The 2026 AI revenue line has to land inside the new $1.5 billion range. And the category ServiceNow is betting on will be settled long before then, and not only by ServiceNow.