ServiceNow's Numbers Say "Buy," Its Chart Says "Not Yet"

# ServiceNow's Numbers Say "Buy," Its Chart Says "Not Yet"

Here's a puzzle for you: a company grows subscription revenue 24.5%, crosses a $1 billion threshold in AI bookings that Wall Street has been begging to see for two years, beats every line item in its guidance, and raises the outlook for good measure — and the stock is down more than 50% year-to-date, sitting on its 52-week low. That's not a typo. That's ServiceNow right now, and it's one of the more jarring disconnects between fundamentals and price action you'll find in large-cap software.

Let's start with what actually happened. ServiceNow's Q2 2026 print, reported in late July, was clean across the board: $3.99 billion in total revenue, up 24% year-over-year, with subscription revenue of $3.877 billion growing 24.5% (23% in constant currency). EPS came in at $0.90 against estimates of $0.86. Management didn't just beat — they raised full-year subscription guidance. And buried in the release was the line that mattered most to anyone who's watched this stock's AI narrative: AI annual contract value blew through $1 billion, finally answering the question that's dogged every enterprise AI story since ChatGPT showed up — is this generating real revenue, or is it just a very expensive demo? For ServiceNow, the answer is now "real revenue," backed by Now Assist adoption and broad-based strength across ITSM, CRM, cybersecurity, and employee workflows.

And the stock got hammered anyway. The immediate reaction to earnings was a 6.5% drop — sellers using strength to exit, not buyers stepping in on the beat. Since then, the bleeding hasn't stopped: shares are trading near $95, below both the 20-day and 50-day moving averages, deep in a technical structure that can only be described as broken. Forward P/E has compressed to 19x with a PEG of 0.92 — numbers that on a standalone basis look like a steal for a company still growing subscription revenue north of 20%. Institutional distribution is active, short float has crept up to 6.08%, and retail sentiment — 100% bullish at these lows, according to the data — is getting run over by whoever's been selling into every bounce.

So which market is right? The one pricing NOW like a compounding AI winner, or the one pricing it like a broken growth story?

My take: this is a textbook quality-derating, and it's exactly the kind of setup that separates people who read income statements from people who trade tape. The fundamentals aren't in question — gross margins at 76.5%, operating margins recovering to 13.3%, $5 billion in trailing free cash flow, no debt-funded desperation here. The AI ACV trajectory toward a rumored $1.5 billion full-year target is a legitimate multi-year growth driver, and the switching-cost moat around enterprise workflow software is as real as it gets. Bulls modeling intrinsic value north of $200 aren't smoking anything exotic — they're just extrapolating durable 20%+ growth against a stock that's been de-rated to a fraction of its five-year median multiple.

But — and this is the part that matters for anyone deciding what to do this week rather than in 2029 — you don't buy a broken chart because the story is good. The $105 level is the line in the sand; reclaim it on volume and this turns into a completely different, much more constructive conversation. Lose $92 and you're staring at a stock that the market has decided doesn't deserve the benefit of the doubt regardless of what the AI ACV number says. The bear case DCF at $73 isn't crazy in a world where hyperscaler competition (Microsoft's Copilot, AWS Bedrock) starts commoditizing the workflow layer ServiceNow has built its premium on.

The setup here isn't "is ServiceNow a good company." It clearly is. The setup is "has the market finished punishing it yet." Right now, the tape says no. When institutional selling exhausts itself and price stops making new lows below $92, that's your signal the derating is over and the fundamentals get to matter again. Until then, this is a stock to watch, not chase — the AI story is real, but so is the distribution.