ServiceNow (NOW) at 84 Times Earnings, Betting on Otto

Subscription revenue grew 24.5% in ServiceNow‘s June quarter, to $3.877 billion, and beat the high end of the company’s own guidance by 150 basis points. At a company this size, that kind of beat does not happen by accident, and it is the number I keep coming back to when people ask whether the AI story here is real or just a slide deck. Current remaining performance obligations, the contracted revenue due in the next twelve months, rose 21% to $13.2 billion. Both numbers came in ahead of what management had told investors to expect three months earlier.

The stock trades around $135.47 as I write this, 30.4% below its 52-week high of $195, at 84.7 times trailing earnings. That is a rich multiple by almost any measure, and it means the stock has very little room for a quarter that merely meets expectations rather than beats them.

My thesis: ServiceNow’s underlying subscription growth is decelerating on paper, from 24.5% this quarter to a guided 20.5% next quarter, but the AI product itself just became more ambitious, not less, when the company opened its platform to competing AI agents rather than only its own. That shift matters more to the next two years of growth than the deceleration in the guided percentage.

The growth rate that keeps landing high

Full-year subscription guidance now sits at $15.76 billion to $15.78 billion, implying growth of about 22.5% for a business already generating more than four billion dollars a quarter. Companies rarely hold anywhere near 20% growth once they cross $15 billion in run-rate subscription revenue; most software peers of comparable scale have settled into low-teens growth by this point. ServiceNow has not, which is the first credit I would give management here. Third-quarter guidance calls for subscription revenue of $3.975 billion to $3.980 billion, up about 20.5%, and cRPO growth of about 20%. That is a real step down from 24.5%, and it is worth taking at face value rather than assuming the usual sandbagging, because guidance has actually tightened, not loosened, over the last few quarters.

Three numbers, one pattern: 24.5% subscription growth, 21% cRPO growth, 22.5% full-year guidance. None of them is decelerating the way a maturing $15 billion software business usually does. That is rare. It is also fragile, because a business growing this fast off this large a base has almost no cover if one quarter slips.

The rebrand that is really a strategy change

ServiceNow renamed its AI product from Now Assist to Otto this year, which sounds like marketing until you look at what came with the name change. The company restructured its AI tiers around maturity rather than just feature count: a foundation level for task assistance, an advanced level for agentic workflows, and a top tier built for autonomous agents that act without a person approving each step. More significant than the name is a licensing shift that began rolling out in July: third-party model providers now sit as the default option for out-of-the-box Otto skills and agents, rather than ServiceNow’s own models exclusively. At its Knowledge conference this year the company went further, opening its full system of action, the workflow and data layer underneath its products, to any AI agent, including ones built by Anthropic or Microsoft, not only agents built on ServiceNow’s own stack. I read that as a bet that ServiceNow’s moat is the workflow data and permissions layer other agents need to act inside a large company, a similar argument to the one I made about Nvidia’s software layer being stickier than its chips, applied here to a company that owns none of the underlying model.

That is a genuine strategic change, not a cosmetic one. Opening the platform to rival agents could have looked like conceding the AI model race. It was not a retreat. Instead, management is betting that owning the system of record and the approval workflow is worth more than owning the model that runs inside it, because a company’s IT and HR data does not move to a new AI agent’s home platform easily once ServiceNow already sits underneath it. The model is replaceable. The workflow data rarely is.

What a customer actually buys

Almost nobody outside a company’s IT or HR department opens ServiceNow directly, and that is by design. Ask an IT director what it does, though, and you get a very different answer than a shrug. It sits underneath help desks, onboarding, approvals, and the routine administrative work that keeps a large organization running, and once it is wired into those processes it is expensive and disruptive to rip out. That switching cost is the reason a 24.5% growth rate is even plausible at this revenue base: existing customers are not just renewing, they are adding modules, and now they are adding AI seats on top of workflows they already run through the platform. The AI pitch is specifically that an agent resolves a ticket or completes an approval without a person doing it manually, which turns the sales conversation into a return-on-labor argument rather than a features argument. That argument is easier to make convincingly than most enterprise software AI pitches, because ServiceNow already has the workflow data to show exactly how many hours a given process used to take.

Where the multiple gets uncomfortable

At 84.7 times trailing earnings against a five-year average closer to 228.1, and 73.3 times forward estimates, the market has priced in a lot of the AI upside already. Forward EPS growth is implied at 16%, meaningfully below the subscription revenue growth rate, which tells me margin expansion is doing less work in the model than top-line growth is. The average analyst price target sits at $142, only 5% above the current price despite 90% of the 29 analysts covering the stock rating it a buy: even bullish analysts are not pricing in much more room to run from here. I ran a version of the reverse DCF exercise I use on expensive growth stocks on this one, and the math is unforgiving: at this multiple, the market needs high-20s percentage growth to persist for years, not the low-20s ServiceNow is now guiding to. Compare that to Microsoft at roughly 27.5 times earnings, a business growing more slowly in percentage terms but off a vastly larger base and at a fraction of ServiceNow’s multiple; the gap tells you how much of ServiceNow’s price is pure growth premium rather than current cash flow.

Metric	ServiceNow (NOW)
Price	$135.47
P/E (TTM)	84.7
P/E (forward)	73.3
Q2 subscription revenue	$3.877B, +24.5%
cRPO	$13.2B, +21%
FY2026 subscription guidance	$15.76B-$15.78B, +22.5%
Q3 2026 subscription guidance	$3.975B-$3.980B, +20.5%
Analyst buy rating	90%
Avg price target upside	5%

Sizing a position at this multiple

A stock priced for high-20s growth carries real downside if a single quarter disappoints, and the post-earnings move here has run as high as an average of 8.5% on results day, though the actual reaction to the July 22 report was a milder -3.7%. That is the kind of volatility profile where the position-sizing framework I use for high-beta names matters more than usual, because a full-size position in a stock that can move nearly nine percent in a session on a growth miss is a different bet than the same dollar amount in a slower-moving business. Short interest is modest at 2.9% of the float, so the skeptics are not making an aggressive bet against the stock even at this valuation; they are mostly on the sidelines rather than pressing a short case. That tells me the disagreement here is less about whether ServiceNow is a good business and more about how many years of 20%-plus growth the current price already assumes.

The risk to this view is straightforward: if net-new annual contract value growth slows and cRPO growth drops meaningfully below the 20% already guided for the third quarter, the market will treat that as evidence the AI upsell is not converting into bookings fast enough to justify the multiple, and a stock priced for high-20s growth does not have much cushion for a guided-down quarter. I would also be wrong if a rival platform, not a rival model, manages to pull workflow data out of ServiceNow’s system faster than I expect; that has not happened at scale yet, but it is the scenario that would break the switching-cost argument this whole thesis leans on.

The number I want next is cRPO growth in the October report. A print at or above the 20% guided would confirm bookings are still outrunning the deceleration story the headline growth rate suggests. A print meaningfully below that, closer to the high teens, would be the first real sign that this quarter’s beat was a peak rather than a new plateau, and I would size down rather than add on weakness in that scenario. Watch that one number. It will say more than the headline growth rate does.

I would also weigh how the Otto rollout shows up in the next two earnings calls, specifically whether management names actual net-new AI seat revenue rather than folding it into the general subscription number. Right now the AI contribution is inferred from the beat, not stated directly, and a company confident in the product usually starts breaking that number out once it gets large enough to matter on its own. If Otto revenue never gets its own line item, that silence would tell its own story.

Analysis and opinion only, not investment advice. Figures come from ServiceNow’s second-quarter 2026 results on SEC EDGAR and its investor site; valuation multiples are approximate and were checked on September 22, 2026.