The headline was fine. The line that broke the stock was buried. ServiceNow fell 27% on Thursday, its worst single session since it listed in 2012, after a quarter in which the company beat on revenue and missed on almost everything investors had been told to watch.
Adjusted earnings of $1.00 a share came in two cents below the consensus estimate. That gap alone would not move a company of ServiceNow’s size. What moved it was the disclosure underneath: subscription bookings and new-contract growth decelerated, and the company’s pattern of beating its own guidance by a comfortable margin did not repeat. Analysts at Citi wrote that the quarter lacked the upside investors have come to expect, which is a polite way of saying the machinery that produced those surprises has slowed.
ServiceNow sells workflow software. Its platform sits above the systems of record that large organizations already run, routing requests between them: an employee needs a laptop, a customer reports an outage, a security team finds a misconfigured server. The company built a business of roughly $15.8 billion in annual subscription revenue by being the layer that makes other software usable, and it has grown at a rate that enterprise software peers have struggled to match.
That model has one exposed flank. ServiceNow charges largely by the seat, and its customers have spent two years being told that AI agents will soon do much of the work those seats perform. If an agent resolves the tickets, the headcount that justified the licenses shrinks. Every seat-based software company faces the same question; ServiceNow faces it with a higher multiple and a more direct exposure, because routing work between systems is precisely the task that agents are being built to perform.
The company’s answer has been to sell the agents itself. It has pushed an AI layer across its platform and reported that the annual contract value of new AI deals rose 65%. It has also built out a forward-deployed engineering function, sending its own people into customer sites to get the products working, a model that consulting firms use and software firms historically avoid because it is expensive and does not scale the way licenses do.
That investment has not yet shown a return, and management said as much by omission: the 65% growth in AI contract value is a striking figure attached to a business whose overall bookings growth is slowing. Either the AI deals are too small to move the total, or they are replacing revenue that would otherwise have been booked as traditional subscriptions. Both readings hurt.
The company did raise its full-year subscription revenue guidance, to $15.78 billion to $15.79 billion from a slightly lower range. That is the detail that makes the selloff interesting. Guidance went up while the stock went down, which means investors were not reacting to the current quarter at all. They were repricing the next several years.
The repricing has been building across enterprise software. Through the past year, the sector’s valuations have compressed as buyers questioned whether per-seat pricing survives contact with agents that can be given a task and left alone. Salesforce, Workday and SAP have all faced versions of the same argument. The difference on Thursday was that ServiceNow, which had been treated as the cleanest growth story in the group, produced a quarter that could be read as confirmation.
There is a counterargument, and it rests on the nature of the work. The tasks ServiceNow coordinates are not creative. They are procedural, cross-departmental and governed by rules that differ from one company to the next. That is exactly where agents are hardest to deploy and most valuable once deployed, because the alternative is a human opening six systems and copying fields between them. If ServiceNow can host those agents and charge for the outcomes rather than the seats, the addressable market grows rather than shrinks.
The risk is in the transition. A company that switches from charging per employee to charging per resolved ticket takes a revenue cut on the way through, and the customers signing AI contracts today are the same ones whose seat licenses renew in 2027 and 2028. Wall Street is willing to fund a transition when the new business is visibly larger than the old one. It is not willing to fund one on a promise.
Chief Executive Bill McDermott has spent his tenure arguing that ServiceNow is a platform company rather than an application company, and the pitch has worked since he took the job in 2019. Thursday’s session put a price on the next phase of that argument. The 27% decline is not a verdict on the quarter. It is the market’s estimate of how much uncertainty now sits between the company’s current revenue base and the one it says is coming.
What would change the picture is disclosure. ServiceNow reports AI contract value as a growth rate, not a dollar figure, and it does not break out how much of its subscription revenue comes from consumption-based pricing. Until those numbers are visible, investors have to choose between trusting the 65% and trusting the bookings line that slowed in the same quarter. On Thursday they chose the line.
