ServiceNow Fell 50%. It Also Just Beat Earnings.

If you glanced at a ServiceNow chart this month without knowing what happened in December, you'd think something had gone badly wrong. The stock, which spent years trading between $700 and $1,000 a share, is sitting at $95. That looks like a company in freefall.

It isn't, quite. ServiceNow did a 5-for-1 stock split in December 2025, its first ever. Rebase the old chart and $95 today is roughly $475 in old-share terms, but the real move is still ugly: down from a post-split high near $210 to about $95-98, a decline of roughly 50% from the 52-week high. It's a stock split story wearing a crash costume, easy to confuse if you're not watching the calendar.

Here's what makes the timing interesting. The day before this piece went to print, on July 22, 2026, ServiceNow reported Q2 earnings: revenue beat, subscription growth accelerated to 24.5% year over year (up from 22% the quarter before, an unusual direction seven years into the company's cloud life), and AI bookings crossed $1 billion in annual contract value. The stock rose 3-4% on the print.

So the question this newsletter actually wants to answer isn't why ServiceNow crashed. It's why a business with accelerating growth is still down 50% from its high, and whether that gap is an opportunity or a warning.

The moat is real, and it's not just marketing

Start with what isn't in dispute. ServiceNow's core product, the "Now Platform," is workflow software: it digitizes how large companies manage IT tickets, HR cases, and customer service requests, all on one shared data model. Once a company builds its incident-management or HR-case workflows on top of it, ripping it out means re-platforming years of process logic, not just swapping a vendor.

You can see that lock-in in the numbers, not just in the pitch. Renewal rate sits above 98%, and net revenue retention, the rate at which existing customers alone expand their spending, runs 120-125% a year, meaning the current customer base grows revenue by a fifth to a quarter annually before a single new logo signs. The biggest accounts, those spending over $20 million a year, are growing fastest of all at 30% year over year: the customers most locked into the platform are the ones spending the most to expand it further.

That's the real bull case, and it doesn't require believing in an AI narrative. ServiceNow's core ITSM (IT service management, the ticketing and incident-response category the company grew out of) business holds an estimated 44.4% share, and Gartner has named it the ITSM Platforms Leader nine years running, according to Gartner-sourced data ServiceNow publishes.

ServiceNow also states that roughly 85% of the S&P 500 runs on the platform in some form, a vendor-reported figure rather than an independently audited one, but directionally consistent with how embedded the company already is inside the largest companies in the world. Renewal and retention are the sharper evidence, though: they're behavioral numbers, describing what customers actually did with their budgets quarter after quarter, while the bear case pricing the stock is almost entirely forward-looking, a bet on what AI agents might do over the next several years.

Net revenue retention of 120-125% means ServiceNow's existing customers alone, with zero new sales, grow the company's revenue by a fifth to a quarter every year.

The fear that's actually pricing the stock

None of that explains the 50% drawdown. What does is a market-wide fear that generative AI agents are about to make platforms like ServiceNow less necessary, not more. The logic: if an AI agent can reason directly over a company's unstructured data and just execute a workflow, why pay for a pre-built, governed layer to do it?

ServiceNow's answer is to lean into governance rather than compete on raw model capability, rather than try to out-build Anthropic or OpenAI on foundation models. It has signed multi-model deals with both Anthropic (for agent-building and regulated workflows) and OpenAI (for multimodal and voice use cases). And it opened "Action Fabric," letting third-party AI agents, including Anthropic's own Claude Cowork, execute actions inside ServiceNow's governed system rather than around it.

CEO Bill McDermott has said this plainly and publicly:

"SaaSpocalypse nonsense." That's how CEO Bill McDermott describes the thesis behind his own stock's 2026 crash.

Whether he's right is genuinely unresolved, not just diplomatically unresolved: this is a real, live architectural bet. The strongest evidence for taking the AI risk seriously isn't the stock price, it's that ServiceNow itself, rather than trying to out-build the most credible AI-native ITSM challenger, Moveworks, simply bought it for $2.9 billion in December 2025. When a company with ServiceNow's engineering resources decides the fastest path past a threat is to acquire it rather than out-code it, that's a tell about how seriously management takes the threat, not a sign it's been neutralized.

The single biggest structural risk, though, isn't a startup. It's Microsoft. Copilot Studio's agent-building tools arrive as an incremental feature inside an M365, Azure, or Dynamics 365 relationship that most large enterprises already pay for, while ServiceNow has to sell Now Assist as a new line item on top of the same budget. That's not a feature gap ServiceNow can close by shipping faster; it's a distribution problem, and distribution problems are the hardest kind to out-execute.

Salesforce is the other name worth watching, for a different reason. Its own AI agent product, Agentforce, has been growing its booked revenue at 169% year over year, faster than ServiceNow's comparable AI numbers, albeit off a smaller base and rooted in CRM rather than IT service management. Every major enterprise software vendor, Microsoft, Salesforce, SAP, Workday, has landed on roughly the same pitch within the same 12 to 18 months ("we are the AI agent orchestration layer"), and being first in the category stopped being a moat the moment everyone else arrived at the same idea. What's left to differentiate on is execution, not positioning, and execution is measured in quarters, not slide decks.

The part ServiceNow did to itself

Here's where the story gets more complicated, and more recent. Layered on top of the AI-disintermediation question is a second, self-inflicted risk that has nothing to do with agents at all.

ServiceNow has historically grown through small, bolt-on acquisitions; in the last thirteen months, it did the opposite, doing two large, back-to-back deals instead. Moveworks closed in December 2025 for $2.9 billion. Armis, a cybersecurity and device-visibility company, closed in 2026 for $7.75 billion, funded through a combination of cash, a term loan, and later a $4 billion bond offering.

The balance sheet effect shows up immediately in the numbers. Total debt roughly tripled, from $2.4 billion to $8.5 billion, in two quarters. The current ratio, current assets divided by current liabilities, a basic measure of near-term liquidity, fell from 1.00 to 0.70, meaning current liabilities now exceed current assets by about $3.6 billion. And GAAP operating margin, the profit margin calculated under standard accounting rules rather than the company's preferred adjusted measure, collapsed from 13.7% for full-year 2025 to just 4% in the quarter that just reported, largely attributed to the Armis deal.

The company's own guidance had flagged roughly 125 basis points (1.25 percentage points) of margin headwind from Armis in the quarter. The actual GAAP margin drop was closer to 8-9 percentage points, most of that difference stripped back out in the non-GAAP figure, which held essentially flat at 29.5%. That gap between GAAP and adjusted profitability is worth sitting with: GAAP net income for the quarter was $298 million; non-GAAP net income was $930 million, a more than 3x difference in the same three months. Stock-based compensation and acquisition costs are real, if non-cash, costs to shareholders, and a wide, persistent gap between the two numbers deserves ongoing scrutiny rather than being waved off as an accounting quirk.

To be clear about what this isn't: none of this is a liquidity crisis. Free cash flow ran $4.6 billion in 2025 and the company is guiding to $5.5-6 billion in 2026, more than enough to service the new debt. The bond offering itself drew about $38 billion in investor demand for the $4 billion raised, which is a credit market vote of confidence, not alarm. But it is a real, recent, and voluntary change in ServiceNow's capital allocation posture, from cash-funded organic growth to debt-funded, large-scale M&A, and it happened at almost exactly the moment the market was already nervous about the business for an unrelated reason.

What you're actually paying for it

Here's the number that should temper any "the stock crashed, it must be cheap now" instinct. Even after a 50% drop, ServiceNow trades at a premium to every named peer on every multiple that matters.

Company	Trailing PE	Price/Sales	EV/EBITDA
ServiceNow	56.8x (GAAP) / 27.2x (non-GAAP)	6.7x	34.2x
Salesforce	19.7x	3.1x	12.7x
Workday	44.0x	3.3x	21.3x
SAP	20.9x	4.1x	13.0x

EV/EBITDA compares a company's total value (market cap plus debt, minus cash) to its earnings before interest, taxes, depreciation, and amortization, a rough proxy for operating cash profit before financing and accounting choices distort it. Same read as P/E: higher means the market is paying more for each dollar of underlying profit.

Even measured on the friendlier non-GAAP earnings basis, ServiceNow's 27.2x price-to-earnings still sits above Salesforce and SAP, and it's priced at two to three times its peers' EV/EBITDA multiples across the board. The honest framing: on a GAAP basis, ServiceNow is now cheaper relative to its own trading history than it has been in years. But "cheap relative to itself" and "cheap relative to what else you could buy" are two different claims, and only the first one is currently true.

The research behind this piece ran a simple three-scenario model over a three-year horizon, tying growth assumptions to the company's own guidance:

Scenario	Growth assumption	Target price	Implied return
Bull	22% (matches FY2026 guidance)	$136.50	+43.0%
Base	16% (moderating growth)	$78.20	-18.1%
Bear	8% (share loss or budget cuts)	$42.10	-55.9%

These are this newsletter's own projections, not analyst estimates or a company target.

One detail worth flagging: Wall Street's current consensus price target sits at $138.84, per stockanalysis.com's tracking of sell-side estimates, almost exactly matching this model's bull case. That's not necessarily wrong, ServiceNow's own guidance does support a 22%-growth bull case, but it's worth noticing that sell-side consensus is effectively treating the optimistic scenario as the expected one, not as an upside case. If you're relying on that consensus target as your margin of safety, you should know what assumption is actually baked into it.

So what do you do with this

Running this through a standard checklist of what makes a good long-term holding: ServiceNow passes on business quality, on operating history, and on free cash flow growth. It fails, cleanly, on two things: leverage just tripled in two quarters, and the valuation still carries a real premium to peers even after the crash. Genuine pricing power, evidenced by that 120-125% net revenue retention, is one of the clearest passes in the whole exercise. Whether the business has favorable long-term prospects, the single most important box on any long-term checklist, comes back a deliberate watch, not a pass: the AI-orchestration bet is coherent and backed by real product moves, but it is not yet proven against foundation-model-native competitors or against Microsoft's bundling advantage.

Put together, that's a company at a moment of genuine uncertainty: not a bargain, and not a resolved story.

That argues against two common reactions. It argues against "the stock is down 50%, that's the bargain," because the valuation, and the leverage picture, don't support that being a free lunch. It also argues against "AI is going to gut SaaS, avoid the whole category," because the growth numbers, the retention numbers, and the earnings beat that landed the day before this went to print, all point the other way.

If you already own it, the case for holding rests on watching two things over the next two or three quarters: whether subscription growth keeps accelerating past 24.5%, and whether GAAP operating margin recovers back toward its pre-Armis 12-14% range. If you're deciding whether to start a position, this is a stock for someone willing to hold through genuine, unresolved uncertainty on both counts, sized accordingly, not a stock for someone looking for a clean, already-de-risked entry.

ServiceNow Fell 50%. It Also Just Beat Earnings.

Key facts

The moat is real, and it's not just marketing

The fear that's actually pricing the stock

The part ServiceNow did to itself

What you're actually paying for it

So what do you do with this

Sources