Datadog, Inc. DDOG

Key data

The business

Datadog sells observability: software that watches other software. A company runs its apps across thousands of cloud servers, and when something slows or breaks, Datadog shows which piece failed, on one screen, with the logs, metrics and traces stitched together. You buy one product, then a second, then a tenth, because it is easier to see everything in one pane than to wire together six vendors. That is the whole moat, the switching cost of the system your engineers stare at all day during an outage. The pricing unit is usage: hosts monitored, gigabytes of logs, spans traced, so the bill climbs on its own as a customer's cloud footprint grows, before any salesperson calls. The newer bet is watching AI, monitoring the large language models and agents companies are now shipping, a fresh category of failures that needs its own dashboards.

The numbers

Growth is reaccelerating, and that is the whole story of the last three quarters. Year-over-year revenue went from +28% in Q3 2025 to +32% in Q1 2026 to +36% in Q2 2026, against a full-year 2025 that grew 28%. The company is getting bigger and speeding up at the same time, which is rare at $4 billion of sales and is what the multiple is paying for.

Read the reported profit with care, because it barely exists. Q2 2026 net income was $45 million on $1.12 billion of sales, a 4% margin, and diluted EPS actually fell from Q1 as share count and stock compensation rose. Over the last year GAAP operating margin was 0.4%, a rounding error. Yet the business threw off about $1.16 billion of free cash in the same twelve months. The gap between those two numbers is stock-based compensation, running at 21% of revenue.

Revenue compounded 35% a year from 2021 to 2025, a triple in four years, and 1H 2026 is running ahead of that pace. The software tension is in the margins below.

The variant this memo holds: the reacceleration to mid-30s growth is real and AI-driven, not a one-time bump, and 19 times sales does not fully price a durable re-steepening. The single print that settles it is Q3 2026 revenue growth, whether it holds above 33% or fades toward the high-20s.

Management

Both founders run it still, Olivier Pomel as CEO and Alexis Le-Quoc as CTO, and both are steady sellers. Insiders sold about $92 million across 83 transactions over the last year and bought nothing, with Pomel's $5.6 million and Le-Quoc's $4.6 million on September 2 the largest. Plan status is not disclosed in the feed, so read these as routine, not signal; $92 million against a $76 billion company is one-tenth of one percent. What deserves the harder look is the pay: stock compensation at 21% of revenue is what turns a 29% cash margin into a 4% reported one, and it is the shareholders, not the income statement, who carry that cost through dilution. No dividend, modest buybacks, R&D at 44% of revenue: this is a company reinvesting nearly everything.

How it fails or surprises you

Growth fades. The 19x sales premium rests entirely on mid-30s growth holding. If Q3 or Q4 2026 year-over-year slips back toward the high-20s where 2025 ran, the AI-bump read dies and the multiple compresses toward the low-teens floor it saw at the 2022 bottom. The tell is one quarter of decelerating revenue, and usage pricing makes that visible fast.

AI watching compounds (right tail). LLM and agent monitoring is a new usage line that bills on top of existing spend, and the market is paying for the core business, not this. If AI observability becomes a second growth engine and lifts spend per customer, growth could hold in the mid-30s for years. First proof: management breaking out AI-product revenue, or gross margin holding near 79% while volume surges.

The profit went backwards. The fact this read explains least: net income fell from $184 million in 2024 to $108 million in 2025 while revenue grew 28%, and GAAP operating income turned negative. Costs outran a fast top line. If operating leverage never arrives and stock comp stays at 21%, the cash story never converts to owner earnings.

Closing thoughts

A specific print settles this one, so name it: Q3 2026 revenue growth. Above roughly 33% and the reacceleration is durable, the AI-watching demand is structural, and 19x sales looks like the low-middle of a fair range for a rare compounder speeding up. In the high-20s and the last three quarters were a pull-forward, leaving a barely-GAAP-profitable name at a full price. An ambiguous mid-30s print that comes with softening billings or deferred revenue tips toward the second read, and the move then is to wait for Q4. The left tail is not solvency, the balance sheet is clean with $435 million cash and manageable convertibles; the left tail is paying a growth multiple for growth that quietly stops. On balance the fatter tail is up while the demand cycle runs, but the downside is a real de-rating, not a scare.

The bet is still that companies keep pouring money into cloud software and keep paying Datadog to watch it, and as they bolt AI into everything, they buy even more watching because AI systems fail in new ways. What breaks it is the growth rate, and the one pair of numbers that tells you first is next quarter's revenue growth against this quarter's 36%, read alongside deferred revenue, which grew only 8% over the last six months and is the softer note under a loud top line.