ServiceNow Sold the AI Flywheel; the Market Wanted the Receipts

A third raise of the AI target and a rebranded 'Superintelligence Control Tower' could not outrun an 18.5% Q4 CRPO guide — arriving just as the stock had already ripped 56% in 90 days

NOW · Earnings Call · 2026-10-28

The beat nobody could use

ServiceNow did the thing it always does. Subscription revenue of $3.988 billion grew 20.5% in constant currency, 50 basis points above the high end of guidance; current RPO of $13.65 billion grew 21%, a 100bp beat; non-GAAP operating margin of 32.5% cleared guidance by 150bp; 19 deals over $5 million in net new ACV, up 135% year over year. Bill McDermott raised the AI ACV target for the third time in a year — from $1 billion to $1.5 billion, and now:

Today, we are raising that target again, this time to $1.7 billion plus by year-end. And as you've come to know me pretty well by now, I've challenged the team to even surpass that number.
Bill McDermott, Chairman and Chief Executive Officer · 2026-10-28

And the stock still fell after hours, because the Q4 guide paired that with current RPO growth of 18.5% in constant currency. Alex Zukin said the quiet part out loud: “all of the metrics that you're citing sound like you're expanding your growth opportunity... Obviously the stock is down pretty meaningfully aftermarket on some of the, I think, the 4Q guide.” — Alex Zukin, Analyst · 2026-10-28 The tape explains the violence. Over the last 90 trading days ServiceNow returned +56.5% in one unbroken leg, printing a period high of $158.96 on October 23 — five days before the print — and sat just 5.6% below it. Meanwhile the full history still shows a 35.9% drawdown from the January 2025 peak of $234.08. This is a stock that had already voted the recovery; the 8-K was asked to ratify it. It didn't. It is also worth noting that this specific debate is not new. In October 2025, Michael Turrin pressed on the fourth-quarter subscription guide of 18% being lower than where cRPO seemed to be settling, and Gina Mastantuono answered that “on-prem is definitely a factor a bit in Q4.” — Gina Mastantuono, President and Chief Financial Officer · 2025-10-29 In July 2026, Tyler Radke asked about the "relatively modest full year subscription revenue raise." Third time, same mechanic, harsher reaction — because the multiple had re-rated.

A new coat of paint on the Control Tower

What genuinely is new is the framing. Superintelligence Control Tower did not exist as a phrase in the prior quarter's vocabulary; the company had been using plain "AI Control Tower" since mid-2025. It now leads the keyword table by a mile, and it carries a bigger idea: ServiceNow as the governed execution layer for enterprise AI, not a workflow vendor with AI bolted on. The supporting datapoints are unusually concrete — assists on Action Fabric up 500% month over month, year-to-date agentic AI consumption in production up 13x, AI deal count above $1 million up 4.5x. The go-to-market logic has also flipped from defense to offense. Where prior calls leaned on partnership announcements as validation, this one used partners as distribution: a ServiceNow plugin inside ChatGPT, and an expansion with Gemini Enterprise where Google supplies the conversational front end and ServiceNow supplies the deterministic rails. McDermott's read on the enterprise mood was blunt — “They all went through the token maxing phase. Here's the big lesson they all learned. All of the vibe slop flops.” — Bill McDermott, Chairman and Chief Executive Officer · 2026-10-28

The part that actually changed the numbers

The most revealing new disclosure was not a product at all; it was an accounting choice. To pull customers into production faster, ServiceNow is now bundling implementation delivery into select large deals, and carving that value out of subscription revenue into professional services — 25 basis points of drag in Q3, roughly 50 basis points expected in Q4. That is the whole reason the Q3 beat did not flow fully into the year, per Mastantuono: “we've made the strategic decision to invest in their success by including delivery expertise with our platform solutions to accelerate AI deployments for select large deals.” — Gina Mastantuono, President and Chief Financial Officer · 2026-10-28 The delivery expertise is real and it works — customers receiving support adopted twice as fast in the first five weeks — but it is also an admission that the consumption flywheel does not yet self-ignite. Someone has to hand-crank it. That shows up in the last filed quarter's economics. Gross margin of 70.7% was down 6.8 points year over year from a 79.2% peak, GAAP operating margin of 11.0% was down 3.7 points, and free cash flow less stock-based comp of $190 million fell 42%. The balance sheet moved too: effective net cash of -$814 million versus +$8.57 billion two quarters earlier. That is what buying Armis, Veza and Moveworks with cash plus buybacks does.

The counter-positioning nobody is paying for yet

Here is the tension worth holding. The market is currently dismantling anything with AI capex attached — AI data center names are down 5.4% over 90 days, with 50 decliners against 6 advancers, and memory and optics themes are worse. ServiceNow is running the opposite pitch: “We're not compute constrained. We don't have capital constraints.” — Bill McDermott, Chairman and Chief Executive Officer · 2026-10-28 Capital expenditure of $114 million fell 40%, and management raised the full-year free cash flow margin target by 100 basis points on the back of it. On paper this is the cleanest AI story in software: revenue up 24%, capex down, no data centers to fund. The market is not paying for it yet — price to revenue of 7.0x is down roughly 61% year over year even as revenue grew 24% to $4.0 billion. That gap is the argument.

Two things management stopped saying

First, security. In the July call, attack surface and Kill Switch were top-five themes, with McDermott openly musing about contending for the top spot in cybersecurity. This quarter both dropped out of the company's top-30 vocabulary entirely; "Armis" even appears on the decel side of the keyword table. The claims survive in the script — security could be "even bigger in a few years than all of ServiceNow is today" — but the center of gravity moved to governance. That is a downgrade in urgency, not ambition. Second, pricing. Brad Zelnick asked directly whether customers want outcome-based pricing. Mastantuono: “our customers are very focused and quite happy with the hybrid model that we have today and are not asking at all for outcome pricing.” — Gina Mastantuono, President and Chief Financial Officer · 2026-10-28 That is consistent with Amit Zavery's July answer that outcome metrics are “very hard to understand or measure and then have to go back and forth in terms of contracts and everything else else. In enterprise software doesn't work.” — Amit Zavery, President, Chief Product Officer and Chief Operating Officer · 2026-07-22 And the tape agrees with them — outcome based pricing is a 360-day decliner. Rare case where management, the sell-side question and the price signal all point the same way. The bear case is simple: a rule-of-57 company now paying professional services to seed token consumption, carrying net debt after buying a security story it has stopped leading with, guiding a decelerating CRPO into a stock that just ran 56%. The bull case is that the flywheel is exactly one quarter from spinning. Both are in the same print.