Datadog Grew 35% and Fell 20% After Its Largest AI Customer Dialed Back Usage
Hypothesis Opus 5 · Research Opus 5 · Writing Opus 5 · Prompt v1.2
Seven companies that store, search and monitor corporate data reported or re-rated inside one week, and the news split them. On 6 August Datadog, which sells software that watches cloud applications for failures, said its largest customer — a leading artificial-intelligence firm on a nine-figure, 17-product contract — renewed but will cut usage from the third quarter. The stock fell 20% in a session despite revenue growing 35.6% to $1.121bn and full-year guidance rising $140m.
The businesses mostly confirm the rally: Palantir grew 93% with net dollar retention of 157%, Fastly posted its fastest growth in four years, Snowflake and MongoDB were accelerating at last print. The valuations mostly do not. Snowflake's trailing price-to-sales has gone from 11.4x in May to 22.8x; Datadog sits at 21x even after the crash, the top of its twelve-month range.
Whether one customer's optimisation is an accident or the template is what the next quarter settles.
| Ticker | Company | Segment | Trend | 30D | 1Y |
|---|---|---|---|---|---|
DDOG | Datadog | Data & Analytics Platforms | 🟢 Cont. Bull | −9.2% | +81.6% |
SNOW | Snowflake | Data & Analytics Platforms | 🟢 Cont. Bull | +26.4% | +72.3% |
MDB | MongoDB | Data Management & Analytics | 🟢 Cont. Bull | +16.6% | +98.4% |
ESTC | Elastic | Data & Analytics Platforms | 🌱 Emerging Bull | +24.7% | +3.5% |
PLTR | Palantir Technologies | AI & Data Intelligence | ⚠️ Emerging Bear | +35.7% | −5.8% |
FSLY | Fastly | Cloud Infrastructure & Platform | 🟢 Cont. Bull | +17.2% | +240.7% |
AMPL | Amplitude | Other | 🌱 Emerging Bull | +22.7% | +0.5% |
12-month price & trend
| Ticker | Mkt cap | P/E | P/E fwd | P/S | P/S fwd | P/GP | P/GP fwd | EV/EBITDA | FCF yld |
|---|---|---|---|---|---|---|---|---|---|
DDOG | $83.3B | 470.5x | 95.8x | 21.0x | 19.0x | 26.4x | 23.9x | 319.5x | 1.4% |
SNOW | $114.5B | n/m | 171.1x | 22.8x | 18.8x | 34.0x | 28.0x | n/m | 1.0% |
MDB | $32.1B | n/m | 65.1x | 12.3x | 10.8x | 17.1x | 15.0x | — | 1.9% |
| Ticker | Mkt cap | P/E | P/E fwd | P/S | P/S fwd | P/GP | P/GP fwd | EV/EBITDA | FCF yld |
|---|---|---|---|---|---|---|---|---|---|
ESTC | $7.8B | 21.2x | 23.2x | 4.5x | 3.9x | 5.9x | 5.1x | 105.7x | 4.1% |
PLTR | $394.9B | 136.5x | 108.2x | 64.2x | 48.6x | 75.7x | 57.3x | 126.8x | 0.9% |
FSLY | $3.6B | n/m | 45.0x | 5.2x | 4.9x | 8.5x | 8.0x | n/m | 1.4% |
| Ticker | Mkt cap | P/E | P/E fwd | P/S | P/S fwd | P/GP | P/GP fwd | EV/EBITDA | FCF yld |
|---|---|---|---|---|---|---|---|---|---|
AMPL | $1.5B | n/m | 164.5x | 4.0x | 3.7x | 5.5x | 5.1x | n/m | 1.9% |
Valuation & fundamentals
Consensus projections
| Ticker | FY2026E | FY2027E | FY2028E | |
|---|---|---|---|---|
DDOG | Revenue | +28.9% | +21.5% | +23.5% |
| EPS | +20.9% | +17.3% | +23.1% | |
SNOW | Revenue | +29.4% | +30.9% | +25.7% |
| EPS | +72.3% | +59.4% | +41.1% | |
MDB | Revenue | +23.1% | +21.6% | +17.9% |
| EPS | +59.1% | +27.1% | +19.6% | |
ESTC | Revenue | +17.6% | +15.0% | +14.5% |
| EPS | +30.3% | +28.2% | +18.8% | |
PLTR | Revenue | +86.1% | +49.3% | +48.2% |
| EPS | +122.1% | +42.4% | +50.3% | |
FSLY | Revenue | +20.6% | +11.9% | +10.6% |
| EPS | +870.1% | +11.5% | +13.1% | |
AMPL | Revenue | +19.4% | +15.8% | +19.8% |
| EPS | −2.0% | +133.9% | +70.5% |
Forward fiscal years only. Blank means no analyst coverage for that year.
The week the bill arrived
On 6 August, Datadog — the New York company whose software monitors cloud applications for crashes, slowdowns and intrusions, and bills by the volume of data ingested rather than by the user — told investors its largest customer had renewed its contract and would then reduce usage starting in the third quarter. Management described it only as a significant artificial-intelligence company on a nine-figure deal using 17 products, and declined to name it on the call; Wall Street widely assumes OpenAI. The shares fell 20.4%, the largest single-day decline in the company's history, exceeding the drop of March 2020.
The quarter underneath was strong. Revenue of $1.121bn grew 35.6% year on year, an acceleration from 32.2%, 29.2% and 28.4% in the three prior quarters, and the full-year revenue midpoint went up $140m to $4.45–4.47bn. The problem is the shape of the next one: third-quarter guidance implies deceleration to roughly 29%. This is the exact hazard of consumption pricing: the same model that lets an AI customer's spend explode lets it be dialled back at renewal.
One week, not one month
The group's month looks like a steady climb only in aggregate. Between 8 and 22 July these seven names were flat to falling — MongoDB down 15%, Palantir down 5.8%. From 22 July to 7 August they averaged +22.4%, more than the entire 30-day figure, and most of that landed in four sessions. Palantir reported on 3 August; Fastly and Amplitude on 5 August; Datadog on 6 August. Palantir's print acted as a rising tide for data and infrastructure peers including Snowflake and MongoDB, neither of which had reported anything. The tidy "rotation out of chips into software" explanation does not survive inspection either: semiconductors rebounded sharply into 30 July, rising alongside software rather than funding it.
The businesses: CONFIRMS
Every member that reported beat and raised. Palantir, the Denver builder of data platforms for defence agencies and large corporations, grew revenue 92.8% to $1.935bn with a 47.1% operating margin, net dollar retention of 157% and U.S. commercial revenue up 149%, and delivered its largest-ever full-year raise. Fastly, a San Francisco content-delivery and edge-security provider, grew 23.3% — its fastest in four years — with gross margin at a record 65.8% versus 51.3% a year earlier. Amplitude, which sells product-analytics software, grew 21.2% to annual recurring revenue of $410m.
The two consumption names that have not yet reported were accelerating at last print: Snowflake, whose Data Cloud consolidates scattered corporate data for analysis, grew 33.5%; MongoDB, whose Atlas database service underpins developer applications, grew 25.2%. Elastic — the search and vector-database company behind Elasticsearch — is the laggard at 16.0% and decelerating, and lost its chief product officer in a resignation effective 17 July.
Datadog's own counter-evidence matters: non-AI revenue growth reached the high-20s, a fifth straight quarter of acceleration from 18%; the $100,000-plus customer base grew to 4,720 from 3,850 and supplies 91% of recurring revenue; agent tool calls rose fourfold sequentially. Bank of America kept a $305 target, arguing one account is being read as a company-wide problem.
The multiples: CONTRADICTS
Growth is real; the re-rating outran it. Snowflake's trailing price-to-sales was 11.4x in May and 18.6x a year ago; it is 22.8x now, 18.8x forward, and on the fairer cross-group measure — price to gross profit, since Snowflake's 66.6% gross margin is the thinnest here against Datadog's 78.6% — it is 33.9x. Datadog is 21.0x trailing sales after the crash, against 16.1x last August and 11.6x in February: the top of its own range. MongoDB has gone 8.3x to 12.3x on 25% growth; Fastly 2.0x to 5.2x, the largest re-rating in the group.
Two exceptions. Elastic trades at 4.49x sales versus 5.12x a year ago, 23.2x forward earnings and a 4.1% free-cash-flow yield — roughly triple Datadog's 1.4% or Snowflake's 1.0%. And Palantir has genuinely de-rated, from 121.6x sales to 64.2x, because revenue nearly doubled while the stock fell 5.6% over twelve months — it remains the most expensive software name by a distance.
One arithmetic warning: Palantir is 62% of the group's combined $638bn market value, and Fastly, a $3.6bn business, supplies roughly 30 of the 57 equal-weighted percentage points of twelve-month gain. Read either way, the "group" is two stories wearing one coat.
The setup
Where it stands — Business momentum broadly confirms the rally; the multiples on the consumption names mostly do not. Would confirm — Datadog's non-AI revenue growth stays in the high-20s in the third quarter, with the customer cut already in guidance. Would invalidate — Snowflake or MongoDB report product-revenue deceleration, signalling optimisation is spreading beyond one Datadog account. Watch next — Snowflake and MongoDB report their July quarters late August–early September; Datadog's Q3 guide implies ~29%. Valuation — Snowflake 22.8x trailing / 18.8x forward sales versus 11.4x in May; Datadog 21.0x / 19.0x versus 11.6x in February.








