Datadog's Billion-Dollar Problem: When Your Best Customer Stops Looking

Datadog's earnings call on Tuesday was a masterclass in the art of managing expectations downward while insisting everything is fine. Revenue hit $1.12 billion in Q2 with 36% year-over-year growth, earnings beat at $0.65 per share, and yet the stock plummeted 15.6%. The culprit: management disclosed that its largest customer—an AI-native powerhouse responsible for an outsized chunk of recurring revenue—has begun reducing usage on the Datadog platform.

This is not a subtle problem. Q3 guidance implies roughly $19 million in new revenue beyond Q2, compared with $115 million added in the prior quarter. That's a six-fold deceleration in growth pace. For a company trading at 20.6x forward revenue multiples while peers languish at 5x, that kind of velocity matters more than a McKinsey deck matters to a board room—which is to say, it matters entirely.

What makes this particularly acute is that Datadog management has already baked maximum caution into Q3 guidance around the customer pullback. This is corporate speak for "we're not hiding anything because we've already accounted for it." CEO Olivier Pomel told analysts on the call that "the company doesn't control what happens with a specific customer," which is true in the way that a surgeon doesn't control whether a patient decides to see another surgeon. Technically accurate. Not especially reassuring.

The concentration risk has always been Datadog's investor anxiety dream. Management states it applies a higher degree of conservatism to its largest customer—a euphemism meaning they know this relationship is fragile enough to warrant caution. The company now counts five AI-native customers contributing over $10 million in annual recurring revenue. That's five. Non-AI-native customers grew in the mid-20s percent range, which sounds reasonable until you realize it means the company's overall growth narrative is increasingly tethered to whether sophisticated AI companies decide to keep paying for Datadog's monitoring infrastructure.

The mathematics are unforgiving. Datadog's valuation thesis rests entirely on 19% projected revenue growth in 2027, meaningfully ahead of the 11% peer average. That premium multiple—the 20.6x revenue multiple that justifies owning Datadog over cheaper infrastructure software names—is predicated on the company maintaining momentum. A major customer cutting usage isn't a rounding error. It's a stress test of the entire investment thesis.

Bank of America has adopted the most generous interpretation available: management built maximum caution into guidance, the concentration overhang is now "de-risked rather than resolved," and the rest of the business remains strong. This is the analyst equivalent of saying "the patient's other organs look fine." It's not wrong. It's just incomplete.

What's instructive here is what Datadog didn't say. Management didn't announce a new $100 million customer. Didn't trumpet accelerating AI adoption across the installed base. Didn't highlight some competitive win that would offset the one customer's exit. Instead, they walked through guidance and hoped the market would focus on the $1.12 billion revenue number and the fact that non-AI customers still grow at healthy rates.

The reality is messier. Datadog built a surveillance software business in an era when observability was a solved problem for most enterprises. The AI wave created a new constituency—compute-hungry companies that need real-time visibility into distributed systems burning expensive infrastructure. That constituency is now learning, apparently, that other vendors exist. Or that internal solutions work fine. Or that the price Datadog charges no longer justifies the value.

For a company that trades on growth premium multiples in a market increasingly skeptical of concentration risk, this is the kind of confession that haunts earnings call transcripts. Not because it reveals immediate distress, but because it forces analysts and investors to do the work they've been avoiding: calculating what growth looks like when your largest customer becomes indifferent.