ServiceNow delivered the earnings beat Wall Street wanted—$3.57 billion in Q4 revenue, EPS of 92 cents adjusted versus 88 cents expected, subscription revenue up 21% year-over-year—yet the market punished it anyway. Shares fell 3% after hours on the earnings announcement, a jarring disconnect that signals something deeper is shifting. The real inflection is happening in the partnerships: ServiceNow announced an expanded deal with Anthropic just seven days after closing its three-year agreement with competitor OpenAI. This one-week dual announcement to rival LLM providers marks a clear pivot away from single-vendor AI standardization toward a multi-model hedge strategy. The market's negative reaction despite a clean beat suggests investors are concerned about either the acquisition spree's cash burn or the strategic message these partnerships send about enterprise AI commoditization.

Here's the divergence that matters right now: ServiceNow posted exactly what investors should want to see. Adjusted EPS hit 92 cents versus the 88-cent consensus. Revenue clocked in at $3.57 billion, beating the expected $3.53 billion. Subscription revenues—the metric that actually matters for a SaaS company—grew 21% year-over-year to $3.47 billion, crushing the StreetAccount estimate of $3.42 billion. The company's backlog indicator, current remaining performance obligations, jumped 25% year-over-year to $12.85 billion. By any traditional earnings measure, this is a beat executed cleanly.

Yet shares fell more than 3% after the bell on Wednesday. This isn't the market being irrational. This is the market telling you something changed about how it's valuing ServiceNow. The question is what.

The earnings story got obscured by the partnership announcements, but not in the way investors usually discount earnings. On the same evening ServiceNow reported results, it announced an expanded partnership with Anthropic to integrate Claude models directly into its platform. That alone might get buried in earnings coverage. But it lands exactly one week after ServiceNow closed a comparable three-year strategic deal with OpenAI to integrate GPT models across its product suite. Two rival LLM providers. One week apart. Same company. This timing isn't accidental.

ServiceNow CFO Gina Mastantuono characterized the company's acquisition strategy in defensive terms to CNBC: "Our acquisitions are 100% not a pivot away from organic growth. They represent an acceleration of it." She was responding to investor concerns that ServiceNow's recent deal blitz—Armis for $7.75 billion in December, the nearly $3 billion Moveworks acquisition that closed this quarter—suggests the company is acquiring to mask slowing organic growth. That's a reasonable investor concern. Enterprise software companies that resort to M&A to juice growth numbers usually signal they can't grow fast enough organically anymore.

But the dual-LLM-provider strategy tells a more nuanced story. ServiceNow is betting that enterprise customers won't standardize on a single AI model provider. This is the opposite of what happened in cloud infrastructure, where enterprises consolidated around AWS, Azure, or Google Cloud. The assumption behind that consolidation was that picking one cloud provider and going deep would give you the best pricing, integration, and support. The AI model market, ServiceNow is signaling, doesn't work that way—yet.

Why? The models still produce different outputs for different tasks. Claude excels at long-form reasoning and customer-facing applications. GPT models dominate coding and complex multi-step workflows. Enterprise customers have workflows that need both. Rather than forcing customers to pick one provider and accept worse performance on use cases where the other model is superior, ServiceNow is building a control plane that lets different models serve different functions. Mastantuono specifically framed the strategy as positioning ServiceNow as an "AI control tower" for businesses.

That's actually a smart positioning if execution holds. But it also signals something investors may have been betting wouldn't happen: model commoditization. If enterprises need both OpenAI and Anthropic models, neither vendor has achieved moat status. That's fine for a customer wanting competitive pricing. It's terrible for investors betting on one model provider achieving marketplace dominance.

The $5 billion share buyback authorization the board approved also suggests management isn't concerned about needing that capital for emergencies, which is curious given the acquisition pace. Either they're confident in cash generation, or they're managing perception around share count amid a stock sell-off.

Look at what actually moved the numbers. Subscription revenue—which is the business that matters—grew 21% to $12.88 billion for full fiscal 2025. That's solid. The Moveworks deal alone is adding 100 basis points of growth attribution in 2026, though at $3 billion acquisition cost that's expensive growth. More concerning: FY2026 guidance of $15.53 to $15.57 billion subscription revenue implies growth deceleration from 21% to somewhere around 20-21% depending on where they land in that range. For an enterprise software company in the AI era, investors may view single-digit deceleration as a signal.

The market's timing on this sell-off is worth parsing. ServiceNow's been a resilient growth story, and beat expectations are usually rewarded. The fact that it wasn't suggests investors were already spooked by something—probably the acquisition spree's financial commitment in a rising-rate environment where software multiples have already compressed. The partnerships may have just crystallized the concern: if ServiceNow needs to acquire AI companies to stay competitive, and is now hedging on model providers anyway, what is the actual AI moat here?

For enterprises, this dual-partnership move is actually useful information. It validates the instinct many CIOs had to resist single-vendor AI lock-in. If ServiceNow—which has massive negotiating leverage—is saying "we need both OpenAI and Anthropic models to serve our customers best," that's enterprise permission to build multi-model applications internally rather than betting the company on one provider.

The real inflection ServiceNow is signaling isn't in the earnings—those are solid but showing normal deceleration for a maturing enterprise platform. The inflection is in the admission that enterprise AI won't consolidate around one model provider. By booking dual partnerships with competing LLM companies within seven days, ServiceNow is saying the market requires multi-model strategies. For decision-makers, this validates building systems that can swap between providers. For investors, this commoditization signal is likely why the market sold despite the beat—single-model vendors just got more vulnerable. For builders, the playbook is clear: design for model portability, not lock-in. Watch the FY2026 earnings call for color on acquisition integration costs and whether the acquisition spree actually accelerates organic growth as Mastantuono claims.