ServiceNow Has Found Its Tollbooth. Nobody Knows What the Toll Is.
ServiceNow Has Found Its Tollbooth. Nobody Knows What the Toll Is.
Like the railroads of the Gilded Age, which grew rich not from shipping their own goods but from charging tolls on everyone else's freight, ServiceNow is positioning itself as the invisible toll collector that every AI agent inside an enterprise must eventually pass through. The pricing model is real. The revenue is real. Whether enterprise buyers understand what they are paying for is another question entirely.
The mechanism is called an assist. ServiceNow meters every AI agent action on its platform and charges customers per assist on top of their existing subscription fees, according to a review by CIO. Large agentic actions consume 150 assists each against 25 for small generative tasks, burning through the assist pool six times faster at heavy AI usage. ServiceNow executives said on an analyst and investor day call that as AI solves more complex workloads, usage rates climb and so does the value the company captures. The harder the work, the higher the bill.
This is not a bug. It is the business model.
At its annual Knowledge conference in Las Vegas on May 5, ServiceNow expanded AI Control Tower, its governance and observability dashboard, with new capabilities to discover, monitor, and shut down AI agents across an enterprise's full technology stack, whether those agents were built on ServiceNow or stitched together from AWS, Google Cloud, Microsoft Azure, SAP, Oracle, or Workday. The platform runs on two decades of enterprise operational data: one hundred billion workflows and seven trillion workflow transactions annually, providing the context engine that makes autonomous agent coordination possible at scale.
The Moveworks acquisition closed in 2025 for $2.85 billion, ServiceNow's largest deal to date. Moveworks is an agentic AI platform, an AI-native search and resolution layer for enterprise work, and its integration deepens ServiceNow's position as the operating system for agentic workflows inside large organizations.
Q1 2026 results showed the model is producing. Subscription revenue came in at $3,671 million, up 22 percent year-over-year, against a base that the prior year's Q1 had established. The company raised its full-year subscription revenue guidance to $15.735–$15.775 billion, representing twenty to twenty-one percent year-over-year growth. Now Assist customers spending over $1 million in annual contract value grew over one hundred thirty percent year-over-year. The company is now processing hundreds of thousands of AI assists. On the analyst and investor day call, a ServiceNow executive described the economics plainly: as AI solves more complex workloads, usage rates climb and so does the value ServiceNow captures.
The company restructured its pricing tiers on April 9, replacing a five-tier legacy structure with three AI-native levels, Foundation, Advanced, and Prime. Now Assist, Moveworks, Workflow Data Fabric, Context Engine, and AI Control Tower are bundled into every tier rather than sold as add-ons. Existing SKUs reach end of sale on July 1, 2026 — after that date, legacy pricing cannot be reinstated. Customers receive a bundled assist pool with their subscription and must purchase top-up charges when they exceed it.
The problem, for customers, not for ServiceNow, is that the number of assists consumed in any given agentic interaction varies. A simple summarization task might consume the base allocation for a period. A complex multi-agent workflow handling a regulatory compliance check across seventeen systems burns through hundreds of assists in minutes. Customers cannot reliably predict their monthly bills before the invoice arrives.
For CIOs and CFOs already under pressure to trim enterprise software spend, this is a specific kind of headache. You sign a subscription contract. You deploy AI agents to automate work. The agents become more capable over time, which is what you wanted. Your assist meter spins faster, which is what you did not budget for. ServiceNow's revenue scales with your AI success, not in spite of it.
It is moving beyond the age of SaaS into a far more granular era where every last detail of a given workload is monitored, metered, and monetized, said Carmi Levy, an independent technology analyst. The model mirrors approaches from SAP and Salesforce, but ServiceNow's depth as an enterprise workflow platform, with deep hooks into IT service management, HR service delivery, and customer operations, gives it a structural advantage in capturing the assist stream at its source.
The analogy to railroads has a well-known ending. The railroad barons built extraordinary wealth extracting tolls from every ton of freight that crossed their tracks. What they did not anticipate was that trucks would eventually run on roads built with public money, commoditizing the freight and breaking the tollbooth economics. Whether ServiceNow faces an analogous disruption, whether the assist meter faces the same regulatory or competitive pressure that eventually cracked railroad pricing, is an open question that the current bull case does not engage.
What is clear is that ServiceNow has made its bet explicit. It is not selling software subscriptions. It is selling the infrastructure that AI agents run through, and it is metering every step.