In the piece on the health of the AI cycle we posed the question that decides the whole story: the pace of infrastructure investment versus the pace of return it generates. Most companies in the market are on the spending side. ServiceNow (NYSE: NOW), which reported this evening, is on the other side - it sells AI tools to enterprise customers and charges money for them. And therefore its report is one of the cleanest tests in the market of the question of whether AI really rolls into revenue.
This evening it delivered a number almost no other company publishes.
About the Company: The Layer That Manages the Work
ServiceNow sells a platform for managing workflows in large organizations. Every internal process that requires a form, an approval and several departments - a new employee who needs a computer and permissions, a fault reported to IT, a procurement request, an inquiry to HR - is scattered in a large organization across systems that do not talk to each other. The platform sits above them and manages the flow.
This is classic SaaS in its strong form: recurring revenue from multi-year subscription contracts; a very high switching cost, since an organization that built its workflows on the platform and trained thousands of employees on it does not replace it easily; and expansion selling, where growth comes mainly from adding modules to existing customers.
In the last two years the company has repositioned itself as the "control tower of AI" - the layer through which organizations manage, secure and measure AI agents, whether their own or others'.
About the Report: Exactly What Was Reported
The company beat the top end of guidance on every metric - both revenue and profitability, per its statement.
Subscription revenue of $3,877 million, growth of 24.5% (23% in constant currency) - a slight acceleration from a pace of 22% in the prior quarter. Total revenue reached $3,987 million, growth of 24%. The non-GAAP operating margin stood at 29%.
The backlog - the metric that matters. At a SaaS company, revenue tells what already happened; the cRPO (contracted obligations to be recognized as revenue in the next 12 months) tells what has already been signed and is about to happen. It reached $13.20 billion, growth of 21%. The total backlog, across the full length of the contracts, stood at $29.0 billion, growth of 21% - and the company attributes this to longer commitments from customers.
And the truly interesting number: AI crossed a billion dollars. Per the company, its AI revenue passed the billion-dollar mark in annual contract value (ACV). In a market where every CEO talks about AI but almost no one separates out the revenue from it, this is one of the few concrete measurement metrics that exist. Alongside it the company reported that AI-agent deployments grew ninefold in nine months.
Customer depth: 123 deals of over a million dollars in new contract value in the quarter - growth of about 40%; and 658 customers with annual contracts of over $5 million, growth of about 23%.
My Angle
A personal opinion of Ilan Abramov - not advice, not a recommendation
Over the past year we almost eulogized all the software companies, and I am not here to say that burial was overdone. I have no doubt AI will wipe out a large part of them - especially those whose entire value was to be a convenient interface layer over a database. A language model does that today at a fraction of the cost.
But precisely out of that a small group was born that does deserve attention: those that manage to translate the change into practice - into business efficiency at the customer. Not to announce AI - to sell it as a result. And when I look at ServiceNow, that is what I see: a billion dollars in annual contract value is not a slide deck, it is a budget organizations signed for.
My feeling about this company is that AI has definitely changed it - but it harnesses it far more than it endangers it. And that is exactly the difference that will separate, in my view, the software companies that survive the decade from those that do not.
Now, having said that - let us look at the number that should worry.
The Guidance - and Here Is the Nuance
For the third quarter the company forecasts subscription revenue of $3,975-3,980 million, growth of 20.5% (20% in constant currency), cRPO growth of 19.5%, and an operating margin of 31%.
For the full year the subscription-revenue guidance was raised to $15,760-15,780 million (growth of 22.5%), with a subscription gross margin of 81%, an operating margin of 31.5% and a free-cash-flow margin of 35%.
Two things are worth noting here, and it is important to say both. First: the forward numbers are lower than the backward numbers - 20.5% expected versus 24.5% reported, and 19.5% expected in cRPO versus 21% reported. Second: the annual guidance raise was modest, from a previous range of $15,735-15,775 million to a range of $15,760-15,780 - meaning a rise of about $25 million at the midpoint, less than a quarter of a percent.
In the background, at the investor day the company held in May, long-term targets were presented: subscription revenue of over $30 billion, 30% of contract value from AI, and meeting the rule of 60 by 2030
- a metric that combines growth pace with profit margin. CEO Bill McDermott noted that the company currently operates at "rule of 56."
The Bull Thesis
Whoever reads it positively will point out that this is one of the few companies in the world providing numerical proof that enterprise AI generates revenue - a billion dollars in annual contract value, and deployments that grew ninefold in nine months. In a market debating whether the investment will pay for itself, this is a rare measurement point.
Beyond that: growth of 24.5% at the scale of a mature company is exceptional; the company beat the top end of guidance on every metric; the total backlog of $29 billion indicates long commitments that reduce risk; and customer depth is expanding rapidly (123 deals over a million dollars, growth of 40%). And finally, the model combines high growth with real profitability - an operating margin of 29%, and a forecast of 31.5% for the year.
The Bear Thesis
Whoever reads it critically will point to a slowing pace. Subscription revenue grew 24.5% in the quarter, but the guidance for next quarter talks about 20.5%. cRPO grew 21%, versus 22.5% in the prior quarter, and the guidance for next quarter is 19.5%. This is a consistent trend of a declining pace - and in companies priced for growth, the growth pace is usually what moves the stock, not the absolute level.
The annual guidance raise was symbolic - about a quarter of a percent - and that is usually a sign that management does not see further acceleration in the second half. Beyond that remains the structural question we wrote about: ServiceNow's historical model prices by number of users, and if AI agents make organizations more efficient, the number of users may fall precisely when the value rises. The company is developing consumption- and outcome-based pricing models, but this is a complex transition not yet proven at scale.
And finally - regardless of the quality of the business - this is a highly valued stock. At such a valuation, even a report that beats guidance can be received with a decline if the forward pace disappoints.
The debate in one line
The bulls see the cleanest proof in the market that enterprise AI generates money: a billion dollars in ACV, a backlog of $29 billion, a beat on every metric and 29% profitability. The bears see a growth pace declining continuously (24.5% today, 20.5% in guidance), a symbolic annual guidance raise, and a pricing model that may be eroded precisely because of the technology the company sells. Both sides lean on the same report.
Summary
ServiceNow's report provides two facts that matter far beyond the stock itself. The first: enterprise AI is already generating real, measurable revenue - a billion dollars in annual contract value is a hard number, not a statement. For those tracking the question of return on the AI-infrastructure investment, this is a significant data point from the demand side.
The second: the forward pace is more moderate than the backward pace. This does not contradict the first - a company can grow 20% and be an excellent business - but it changes the question the investor should ask, from "is it growing" to "at what pace, and at what price." Both facts are in the same report, and they do not contradict each other. How to weigh them - that is already each reader's decision.
Sources: ServiceNow's official results report for the second quarter of 2026 as filed with the SEC (Form 8-K, exhibit 99.1, July 22, 2026), including the guidance tables and non-GAAP reconciliations attached to it. The chart is shown in real time via TradingView.