Three-Quarters of Datadog's Biggest Customers Now Use Its AI Products

Three of every four Datadog customers paying $100,000 or more a year now use its AI products, CEO Olivier Pomel told investors, and the company’s first specialized model, Scout 1.0, now runs its incident-investigation agent.

Datadog expects fourth-quarter revenue of $1.19 billion to $1.20 billion, level with the third quarter’s $1.20 billion, even as AI adoption spreads. Two things sit behind that forecast: it counts the largest customer at its minimum commitment, and several large AI customers renewed early at higher commitments in exchange for better volume pricing, which management called a near-term revenue headwind.

What is Scout?

Scout 1.0 is built to investigate production alerts. It powers Bits Investigation, Datadog’s AI agent for site reliability engineering, the work of keeping services running. Datadog introduced Scout on Oct. 15 and still lists it as a preview.

On Datadog’s own benchmark of real alerts, Scout at medium effort scored about the same as GPT-6.1 Sol and Claude Opus 5.5 while costing 3 to 8 times less per investigation, by Datadog’s estimate. Datadog did not publish a price for Scout; its cost figures are relative to its own medium setting.

Who gets the savings?

Each investigation takes repeated model calls, so a cheaper model lets Bits Investigation run on more alerts. Pomel said Datadog charges extra for Bits AI, its flagship agent, and keeps that price close to what inference costs it, so savings from more efficient models pass to customers.

Pomel gave the reason: models are improving so fast that a fixed price makes little sense when the volume of work is expected to rise by orders of magnitude while its price falls by orders of magnitude.

Where is the revenue going?

Datadog’s largest customer cut its usage in the third quarter but stayed above its minimum commitment, CFO David Obstler said, and the fourth-quarter forecast counts that customer at the minimum. The early AI renewals came at higher commitments in exchange for better volume pricing, which Pomel described as a short-term revenue headwind.

Pomel said AI adoption, coding agents especially, outran customer budgets. Datadog’s cost tools track the result: more than one-third of its Cloud Cost Management customers now monitor AI costs, and the number doing so is up more than 11 times.

Pomel’s bet is on volume. “As long as we get more usage, deliver more value, we’ll be fine,” he told analysts.