Datadog reported second-quarter results before the open on August 6. Revenue of $1.12 billion grew 36% year on year, adjusted earnings of $0.65 beat the $0.58 consensus, free cash flow held a 25% margin, and management raised full-year guidance to $4.45–4.47 billion. The shares fell 19.0% to $229.29 on more than three times average volume, one session after setting an all-time high of $292.72.

Three reasons the fall makes sense:

1. Datadog disclosed that it renewed its largest customer, a leading AI lab, on a nine-figure deal, and that the customer’s usage will decline from the third quarter. Third-quarter revenue is guided to $1.135–1.145 billion, only 1.7% above the June quarter against the 11.4% sequential growth just delivered.
2. The implied fourth-quarter exit growth rate is roughly 25%, against the 36% just printed. Consensus estimates for 2027 have to come down, and at 17.5 times forward sales the shares had been priced as though they would not.
3. Management declined to say whether the decline reflects churn, in-sourcing or a lower unit price. That question stays unanswered for ninety days, and uncertainty of that shape carries its own discount.

Three reasons the reaction looks overdone:

1. Chief Executive Olivier Pomel told the call that the largest customer contributed nothing to second-quarter growth. The 36% therefore came entirely from the rest of the base, which has now accelerated for five consecutive quarters.
2. Growth among non-AI customers reached the high-20% range, up from the mid-20s last quarter and 18% a year earlier. Customers running ten or more Datadog products rose to 13% from 7%, and remaining performance obligations grew 43% against revenue growth of 36%.
3. Pomel said the company chose to “fully de-risk” the account inside guidance rather than model it. If that is accurate, the third-quarter number embeds close to a worst case rather than an expectation.

Working back from the guidance bridge, the review estimates the account is stepping down by roughly a quarter to two-fifths sequentially, worth $140–190 million annualised, or about 3–4% of the company. That explains a repricing of perhaps 5–8%. The remainder is multiple compression.

The rating stays at Watchlist, but the reasoning has inverted: in July the business was excellent and the price was wrong, and today the price is close to fair with one open question. The preferred accumulation band is $195–220.

Scenario analysis, scorecard and technical levels are set out in the complete Datadog (DDOG) report on our Reports page.