ServiceNow's Record Drop Misreads Where Enterprise AI Value Will Sit

ServiceNow had the worst day in its history as a public company this week, and the reason had little to do with whether customers are buying its AI. Investors sold the stock because one forward-looking bookings number came in half a point light, and because they fear AI agents will hollow out traditional enterprise software. We think that fear misreads where value will sit as agents spread. Enterprise leaders should not take their platform cues from a stock chart.

What happened to ServiceNow this week?

ServiceNow reported third-quarter results on October 28. Subscription revenue rose 21% to $3.99 billion, the company raised its full-year subscription and free cash flow targets, and it lifted its 2026 goal for AI annual contract value to $1.7 billion, up from $1 billion at the start of the year, according to its earnings release. Chief Financial Officer Gina Mastantuono said production use of agentic AI was up more than 13 times from a year earlier.

The problem was guidance for current remaining performance obligations, a measure of contracted revenue due over the next year. ServiceNow forecast 18.5% growth for the December quarter, below the 19% analysts expected, The Next Web reported. The shares closed down 27% on October 29, their largest drop on record, according to The National. Coverage tied the selloff to "SaaSpocalypse" fears, the idea that AI agents will make seat-based enterprise applications obsolete.

Why the bear case deserves a hearing

The skeptics are not foolish. If agents do the work that employees once did inside an application, a vendor that charges per user loses its pricing base. AI-native companies are building agents that sit above any single system of record. And ServiceNow's own numbers carried warning signs: BMO's Keith Bachman said the outlook "still points to deceleration," and Citi's Tyler Radke noted that investors have yet to see returns from the company's forward-deployed engineering push, according to The Next Web.

William Blair's Arjun Bhatia pointed to accounting effects, according to The Next Web: early renewals pulled bookings into September, some public-sector customers shifted to on-premises deals recognized differently, and more contracts carry termination-for-convenience clauses. Reasonable people can disagree about whether that adds up to a slowdown or a timing blip. What the numbers do not show is customers abandoning the platform.

Why we think the market has it backward

The agent era does not reduce the need for a governing layer. It multiplies it. Every enterprise running agents at scale has to answer the same questions: which agents exist, what data and systems each one can touch, who approved it, what it costs, and how to shut it off when something goes wrong. Those are workflow, identity and audit problems, and they get harder, not easier, as the number of agents grows.

Practitioners at AI Connect events have told us this repeatedly. Hemakumar Velugula of Xnode described a CIO whose pilots stalled for a reason that had nothing to do with model quality. "I'm not worried about proving the technology, but I'm unable to prove I'm in control of it," Velugula recalled the CIO saying at The AI Enterprise Conference 2026. We covered that session in our piece on why the AI stack needs a control plane for agents. At COLLIDE 2026, IBM's Ann Leach described the agent sprawl that follows when each team builds on a different platform. Both speakers came from companies that sell agent governance tools, so their views are not disinterested, but the problem they describe is one enterprise buyers raise on their own.

ServiceNow is betting on exactly this need. On the same day as its results, it launched a Control Tower Alliance with founding members including Anthropic, AWS, Cisco, NVIDIA, Okta, OpenAI and Stripe, built around shared standards for agent identity and guardrails. Okta chief product officer Ely Kahn said in the announcement that adoption stalls when companies cannot centrally revoke a rogue agent's access. The alliance is days old and has no track record, and an industry group whose members include the model makers whose agents it would govern deserves scrutiny until its standards are published and independently tested. Still, whoever solves that problem well will earn durable revenue.

What incumbents still have to prove

Our argument is not that every incumbent wins. It is that the control layer will be valuable, and incumbents have to earn it. Two tests matter.

First, control has to reach across vendors. Leach put it sharply at COLLIDE 2026: "If the control works only inside one ecosystem, it's not really a control plane." A governance product that only sees agents built on its own platform will not satisfy a CIO running agents from three clouds and two model labs.

Second, pricing has to follow the work. Per-seat models fit poorly when much of the work is done by software. Vendors that move toward consumption or outcome-based pricing, with clear cost visibility, will hold customers. Those that defend seats at all costs will invite exactly the disruption investors fear.

What enterprise leaders should take from it

Stock moves measure expectations, not product value. A 27% drop in a day says more about how crowded a trade was than about whether a platform belongs in your architecture.

The better questions are concrete. Can this vendor show you every agent operating in your environment, including ones built elsewhere? Can it enforce policy at the moment an agent acts, not just on paper, the point Kindo's Shan Zhou made in his COLLIDE 2026 talk on runtime AI governance? Can it tell you what each automated workflow costs and returns? Can it cut off an agent in seconds?

Vendors that answer yes will matter more in an agentic enterprise, whatever their share price did this week. Those that cannot will lose ground.

Frequently asked questions

Why did ServiceNow stock fall 27%?

Its forecast for current remaining performance obligations growth, 18.5% for the December quarter, came in below the 19% analysts expected, reviving fears that AI will disrupt enterprise software.

What is the SaaSpocalypse?

It is a market term for the fear that AI agents will make seat-based enterprise software applications less valuable or obsolete.

What is ServiceNow's Control Tower Alliance?

Launched October 28, 2026, it is a partner ecosystem including Anthropic, AWS, NVIDIA, Okta and OpenAI built around shared standards for governing enterprise AI agents.

Opinion pieces reflect the views of The AI Enterprise Magazine Team.

Filed under: Agentic AI, AI Governance & Trust