Front page
NewsService Operations5h · 12:01 GMT · 11 min read
Is ServiceNow’s Billion-Dollar AI Bet Actually Paying Off?
ServiceNow’s AI business has crossed $1 billion in annual contract value, while its latest ITSM tools charge for individual investigations. With Q3 results due October 28, service and contact-center buyers need to know whether that spending buys lasting resolutions and more reliable operations.
We’ve been following service management news for months, and there’s one line from ServiceNow’s July earnings call I want to come back to. CFO Gina Mastantuono told analysts:
“Customers aren’t paying us for tokens, they’re paying for resolutions.”
So clearly, the receipts are becoming more valuable.
ServiceNow had already crossed $1 billion in AI annual contract value by July. October brought Flow, a simpler AI service desk, and a striking customer result from Raleigh, North Carolina: a reported 66% reduction in IT service desk costs, based on staff time redirected to other work rather than verified cash savings.
Impressive figures. Still, a handful of customer success stories can’t tell us how the wider market’s getting on.
We’ve followed ServiceNow’s agentic ambitions since Knowledge 2026. The October 28 earnings call should give us something more substantial to test.
For service leaders planning next year’s budgets, we’re particularly interested in how much customers are spending on agentic ITSM, what they’re getting for it, and whether management’s answers have become any more specific since July.
What Do ServiceNow’s Q3 Results Need to Show?
ServiceNow set itself a demanding benchmark in July. Subscription revenue reached $3.877 billion, up 24.5%, while current remaining performance obligations (cRPO), revenue contracted for the next 12 months, hit $13.2 billion. It also closed 123 deals exceeding $1 million in net new annual contract value, nearly 40% more than a year earlier.
There was plenty of appetite for traditional ITSM, too. CFO Gina Mastantuono told analysts:
“ITSM was in 15 of our top 20 deals.”
On October 19, ServiceNow announced a Leader position in IDC’s AI-enabled ITSM assessment. That supports its standing, although it says nothing about autonomous resolutions. Its Q2 renewal rate was 98% by annual contract value, which also tells us nothing about renewal of individual AI products.
For Q3, management forecast subscription revenue of $3.975 billion to $3.980 billion, approximately 20.5% annual growth. Federal revenue brought forward into Q2, and an expected $35 million currency headwind to Q3 cRPO complicate comparisons. We wouldn’t blame any slowdown on AI without accounting for both.
I’ve got a few questions. ServiceNow passed $1 billion in companywide AI annual contract value, but how much comes from L1 ITSM? More than 40 L1 customers were cited in July, but what’s the absolute scale today?
Internally, its specialist resolved 46% of the 15% of incidents eligible for automation. How many stayed closed? And with premium pricing uplifts and Incident Analyzer consuming four assists per run, what’s the fully loaded cost per successful resolution?
Is Agentic ITSM Actually Driving ServiceNow’s AI Growth?
ServiceNow’s AI business is growing, but the contribution from agentic ITSM is harder to pin down. In July, the company reported a ninefold increase in customers running agentic AI in production over nine months. Impressive, although management didn’t disclose the starting number or how many were specifically using L1 ITSM automation.
Goldman Sachs analyst Gabriela Borges asked the question we’d have asked: How much more are customers spending when they adopt ServiceNow’s Level 1 ITSM automation?
Amit Zavery, ServiceNow’s president and COO, said more than 40 customers were already using its L1 AI specialists, with some deployments completing:
“close to 80%-85% of the service requests without having any human interaction.”
That’s worth investigating. ServiceNow’s own internal deployment paints a more complicated picture: only 15% of IT incidents were eligible for its L1 specialist, which resolved 46% of those eligible incidents. That works out to roughly 6.9% of all incidents. The published case study also reports resolutions 99% faster than human agents.
Different workloads explain some of that variation. Still, we’d want to know exactly which requests qualified before getting too excited about autonomous resolution percentages. Neither result proves what a typical customer’s share of successfully resolved incidents will be.
Borges also asked about incremental customer spending. Management discussed growing AI consumption without providing an L1-specific revenue figure.
That unanswered question deserves another airing, especially when ServiceNow’s next growth story depends on customers trusting AI to finish the job.
How Is ServiceNow Monetizing AI Agents: Licenses, Consumption or Bundles?
ServiceNow <a href="https://cxtoday.com/the-ultimate-enterprise-guide-to-ai-automation-in-customer-experience” rel=”nofollow noopener” target=”_blank”>AI monetization combines premium subscriptions with consumption-based charges. In July, CEO Bill McDermott described its approach as:
“The hybrid with the seat and the consumption.”
CFO Gina Mastantuono said Pro Plus upgrades generated pricing increases above 30%, while newer AI-native packages delivered 20–30% uplifts. Those are substantial product-tier increases, although they don’t establish an individual customer’s total bill or return.
On October 1, ServiceNow introduced Flow, a conversational service desk for Slack and Teams, with consumption-based access for eligible AI-native customers. It launched in controlled availability, with broader North American and EMEA availability expected in Q4. Crucially, new customers can adopt Flow without buying ServiceNow’s wider platform.
That opens a rather awkward commercial question. Is Flow attracting new ITSM buyers, or giving established customers a simpler alternative to the platform they already pay for? Either way, we’d want an actual usage bill before declaring a one-day deployment cheap.
ServiceNow’s Otto for ITSM release notes offer a more revealing detail. Incident Analyzer examines incidents, suggests likely causes and configuration items, and records findings in work notes. It ships switched off, defaults to P1/P2 incidents when enabled, and costs four assists per investigation. An investigation isn’t a completed repair. We’d want to know how often that paid work produces a lasting fix, and who pays for unsuccessful runs.
ServiceNow’s October 8 research found only 28% of 550 executives could attribute nearly all AI spending to specific teams or use cases.
Its October 15 AI Cost Management announcement promises to connect spending with outcomes, but availability isn’t expected until Q4.
We’ve already questioned whether AI service savings justify full operating costs. There’s plenty of enthusiasm for charging according to usage. We’d like more evidence that customers understand what they’re paying for.
More from the Service Operations desk
News
Grafana’s Grades, Contact Center Handoffs and the Carrier Test
At ObservabilityCON, Grafana introduced new investigation tools and previewed ways to link user experience with service ownership. Zoom and Genesys are tightening contact center handoffs, while fresh carrier incidents test whether suppliers can tell when customers actually get through.
Rebekah Carter · 24 Oct 2026News
PSTN Switch Off 2027: 15 Weeks Left to Find the Analogue Line
The UK’s PSTN switch-off is just 15 weeks away, but some businesses face earlier supplier deadlines and higher charges already. Contact centers need to track down forgotten call routes and connected devices, then test every replacement before a customer depends on it.
