ServiceNow just dropped 18% in after-hours trading after quarterly results that beat its own targets but mostly fell short of what Wall Street wanted.
What the numbers actually say
Revenue grew 20.5% year-over-year to $4.11 billion, landing $20 million above analyst consensus. Adjusted earnings came in at $1 per share against an expected $1.02. The earnings miss is not catastrophic on its own. The sales forecast for next quarter also fell short: ServiceNow said $4.25 billion to $4.26 billion; analysts had expected $4.36 billion.
The AI numbers underneath are moving fast. The total value of AI contracts the company expects to collect over a year — its annual contract value, or ACV — is now projected at $1.7 billion for the full year, up 70% from January. CEO Bill McDermott said new AI ACV signed in the quarter rose 65% year-over-year. ServiceNow says EmployeeWorks, an AI-powered employee portal with a search-and-automation chatbot called Otto, reached $100 million in ACV within three quarters of launch. The company's Super Intelligence Control Tower — a platform that maps and monitors every AI model and agent running inside a customer's systems — is growing a core usage metric at 500% month-over-month, McDermott said. McDermott told investors he expects the AI business to reach "true acceleration" next year.
Large enterprise demand is holding. ServiceNow closed 19 deals with annual contract values above $5 million, a 135% increase from a year earlier. A new OpenAI partnership will let customers reach the ServiceNow platform directly through ChatGPT, and a consortium that includes Nvidia will build joint offerings around the Control Tower.
The gap markets are pricing
This is the same pressure visible across enterprise software right now. Strong AI metrics and healthy top-line growth are no longer enough to protect a valuation when the next quarter's revenue number disappoints. It rhymes with the repricing Salesforce faced when its AI agent story outran near-term seat growth, and with the pattern that hit cloud infrastructure names when consumption revenue came in below model. Markets have moved from rewarding AI momentum to demanding that momentum show up as contracted, predictable revenue — and they reprice hard when it doesn't.
The ServiceNow case makes the tension plain. A platform that can watch every AI agent running inside a large enterprise, track the quality of its responses, and measure the productivity it actually delivers is exactly what buyers need as agent deployments multiply. Usage is surging. The question the market is now asking out loud is how long before surging usage becomes surging revenue.
That question matters to anyone running AI at scale. Usage metrics and productivity gains are real, but they do not automatically convert into the kind of contracted, recurring revenue that holds a budget line through a down quarter. Operators who can close that loop — who can show that AI usage produces measurable output and tie that output to a renewal conversation — are the ones who avoid the same gap ServiceNow is now being asked to explain. The stakes are whether AI becomes a durable line item or stays a promising experiment waiting for its number.
