Datadog, Inc. keeps showing up wherever modern software breaks

The Datadog vibe shift

If you’ve ever watched a team ship something “small” and then spend the next six hours in a group chat arguing about whether the outage was the database, the network, the cloud provider, or “some weird edge case,” you already understand why Datadog, Inc. (DDOG) exists.

Datadog’s core pitch is simple: modern software is messy, and when it breaks, you want one place to see what happened—across infrastructure, apps, logs, user experience, and increasingly, security. The company has spent the last few years turning that “single pane of glass” idea into a platform that tries to be unavoidable once it’s embedded.

And heading into February, Datadog is back in the spotlight for a very basic market reason: it’s about to report.

What’s on deck: February 10

On January 20, 2026, Datadog announced it will report fourth-quarter and full-year 2025 results before the U.S. market opens on Tuesday, February 10, 2026, followed by an 8:00 a.m. ET conference call.

That date matters because the market’s current obsession isn’t just “AI winners.” It’s “AI winners who can prove the demand is real, recurring, and not a one-quarter sugar rush.” Datadog sits right at that crossroads: it’s not selling GPU chips or consumer chatbots. It’s selling the picks-and-shovels layer for companies running big, complicated systems—and now those systems increasingly include AI workloads.

The last real checkpoint: Q3 2025’s numbers

Datadog’s most recent reported quarter (third quarter of 2025, reported in early November 2025) did a good job of reminding investors what “durable software growth” looks like when it’s working.

For Q3 2025, Datadog reported revenue of $886 million, up 28% year over year. It also reported non-GAAP operating income of $207 million (a 23% margin) and free cash flow of $214 million (a 24% margin).

The customer story was just as important as the top-line number. Datadog said customers with at least $100,000 in annual recurring revenue reached about 4,060, up 16% year over year in Q3 2025. In plain English: bigger organizations kept leaning in, not backing away.

Then there’s the sticky part. Datadog said trailing-12-month net revenue retention was 120% in Q3 2025. That’s the “customers don’t just stay, they expand” metric, and it’s a big reason this company keeps getting treated as more than a simple monitoring tool.

Why the AI angle is real (and why it’s tricky)

Datadog’s AI narrative isn’t about slapping a chatbot onto dashboards. It’s about the fact that AI workloads create more complexity: new services, new endpoints, more data, more failure modes, and more pressure to keep latency and reliability tight.

In the Q3 2025 earnings call, Datadog highlighted that “AI-native” customers (excluding its largest customer) represented 12% of revenue in Q3 2025, up from 11% in Q2 2025 and about 6% a year earlier. That’s a clean signal that AI infrastructure is becoming a meaningful part of the business.

The tricky part is that management also framed this bucket as something that will matter less over time as “AI usage in production broadens.” Translation: eventually, AI stops being a special category and just becomes part of “normal software.” That’s a good long-term sign for Datadog, but it can complicate the market’s appetite for simple, meme-friendly AI metrics.

What investors are actually listening for in Q4

When Datadog reports Q4 and full-year 2025 results on February 10, 2026, the market will likely care less about any one feature announcement and more about whether the company can keep doing three things at once:

• Keep revenue growth healthy without relying on a single mega-customer
• Keep larger customers expanding (not just signing up)
• Keep platform breadth translating into real usage, not “nice-to-have” shelfware

Datadog’s business is built around the idea that software incidents aren’t rare events anymore—they’re a normal tax on shipping fast. If that remains true (and it does), the question becomes how big Datadog can get as the default system for understanding what’s happening across increasingly chaotic stacks.