ServiceNow Has the AI References; What It Lacks Is the Revenue Line to Match

ServiceNow closed the books on a quarter that offered something for both camps — and satisfied neither. Subscription revenue for the third quarter of 2026 came in at $3,988 million, up 21% year over year, while total revenue reached $4,106 million. Management lifted its full-year subscription revenue outlook to $15.78–15.79 billion and raised the free cash flow margin target to 36%. The market's answer, delivered Thursday, was to mark the stock down sharply while software names more broadly came under pressure on worries about AI-driven disruption.

The sticking point was the adjusted earnings per share, which landed just shy of expectations. That miss, combined with a softer outlook for the closing quarter, outweighed the improved annual targets — a sequence Reuters reported had already shaped the reception of the results. At least twelve brokerages trimmed their price targets on the stock following the report.

Raymond James Wants the AI Story in the Numbers

Among the skeptics, Raymond James singled out the pace of growth in subscription revenue and backlog as falling short. In the firm's reading, ServiceNow's AI initiatives have yet to show up meaningfully in the business — and loftier full-year goals did little to quiet those doubts.

The company, for its part, raised its target for annualized AI contract value to $1.7 billion. That figure is an ambition, not reported revenue, and the gap between the two is precisely where the debate now sits: management is betting on a larger AI contribution, while critics are demanding visible evidence in revenue and bookings.

Customer Wins Put Flesh on the Strategy

Concrete deployments offer the more persuasive case. On October 22, ServiceNow pointed to customer Thrive, which used its Technology Provider Service Management solution alongside AI agents to reach a 73% first-contact resolution rate while cutting time-to-resolution by 63%. Operational numbers of that kind say more about the AI strategy than the sheer size of a partner network — they describe a measurable benefit, with issues resolved more often on first contact and faster overall.

The evidentiary value has limits, though. The figures come from a single customer and from ServiceNow's own communications. They demonstrate neither broad-based success nor, directly, higher profit. Anyone reading a financial breakthrough into them is skipping the decisive proof.

Two further customer commitments arrived on Wednesday. Hewlett Packard Enterprise selected ServiceNow to redesign customer support and field service within its Cloud & AI business, and Stagwell chose the platform to consolidate four service desks spanning more than 70 agencies and 10,000 employees. The Stagwell task is notable for its concreteness — consolidation of that scale offers a real basis for judging the platform's usefulness. But a selection decision is not a documented outcome, and customer decisions belong in a different evidentiary category from customer results.

Guardrails Are the Other Half of the Bet

Governing AI agents is part of the same commercial test. ServiceNow unveiled the Control Tower Alliance on Wednesday, with Anthropic, AWS, Cisco, NVIDIA and OpenAI among its founding members, aimed at helping enterprises deploy and control AI. A day earlier, ServiceNow and NVIDIA announced general availability of ServiceNow Arc, a personal enterprise agent with controls and enforcement features built around NVIDIA OpenShell. The logic holds together: an enterprise agent's value shouldn't be judged apart from its governability, so a credible strategy must offer not just capabilities but rules for their use. Whether customers will fund that layer with fresh budget is a question the alliances themselves cannot answer.

The partnership push extends further. On Tuesday, ServiceNow and Google Cloud expanded their collaboration to connect industry workflows with Gemini Enterprise. Such announcements map the breadth of the AI build-out but say nothing on their own about the incremental revenue or backlog it will generate.

Two Yardsticks, One Stock

It would be too easy to pin the skepticism entirely on ServiceNow, given the wider pressure on software stocks from fears of AI displacement. The reverse inference is no more convincing — not every weakness is a symptom of sector-wide anxiety. The adjusted EPS miss remains a company-specific black mark.

What investors are left weighing is a pair of developments moving in parallel: upgraded annual targets on one side, questions about how quickly AI converts into revenue on the other. The next convincing datapoint won't be another marquee partner. It will be broader evidence that customer benefit and profitable growth are arriving together — and the raised ambitions will be measured against exactly that standard.