Datadog Posts 36% Quarterly Revenue Growth and Says Its Largest Customer Will Cut Usage; Shares Fell 17% on Aug. 6, Rose About 10% on Aug. 10
Datadog beat estimates and raised its full-year outlook, but a warning about its biggest customer sent the stock to its largest one-day drop on record before a partial rebound four days later.
The Number That Should Have Been The Whole Story
Datadog brought in $1.12 billion in the second quarter of 2026, up 36% from a year earlier[1]. That beat what Wall Street expected, and it beat the company's own forecast[4]. Non-GAAP earnings per share came in at $0.65, well above the roughly $0.58 analysts had penciled in[4]. Datadog also raised its full-year revenue guidance, to a range of $4.45 billion to $4.47 billion[1].
By almost any normal reading, that's a great quarter. And yet on August 6, 2026, the stock fell somewhere between 17% and 19%, with some measures showing an intraday drop as steep as 20% to 21%[4][5][16]. It was the largest single-day decline in Datadog's history, worse than the roughly 17.8% drop the stock took in March 2020[4][16].
Four days later, on Monday, August 10, the stock climbed back about 10% to 11%[10][11]. Both of those numbers are real. They happened on different days, for different reasons, and putting them together the wrong way is exactly how this story gets told wrong.
One Sentence, One Customer, One Reversal
The reason for the crash wasn't in the revenue line. It was a single disclosure on the earnings call: Datadog's largest customer, which the company has never named, is going to cut how much it uses the service starting in the third quarter[7][8].
That customer isn't leaving. It just renewed a contract worth more than $100 million, covering 17 different Datadog products[8]. Several analysts believe the customer is OpenAI, but that's their inference, not something Datadog has confirmed[7]. The company has never disclosed which customer it is or how much of its revenue that customer represents, which is part of why the news landed so hard — investors had no way to size the risk until it moved.
The mechanism matters here. Datadog is a monitoring company: it helps other businesses track their software and computing systems, and it charges mostly by volume — how much data a customer sends it to watch. During the recent AI spending boom, that pricing model worked in Datadog's favor. As AI companies scaled up their computing use, their monitoring bills scaled up automatically, with no extra sales effort required.
That same mechanism runs in reverse just as easily. A customer can keep every product, stay under contract, and still slash its bill if it simply sends less data through the system. That's the scenario Datadog described. Third-quarter guidance of $1.135 billion to $1.145 billion implies growth of 28% to 29%, down from the 36% just reported[15].
What Nobody Disputes, And What The Filing Doesn't Say Out Loud
Set aside interpretation for a moment. A few facts here aren't in dispute by either bulls or bears. Datadog's customer base with contracts over $100,000 a year grew to about 4,720, up from roughly 3,850 the year before[1]. Growth from customers outside the AI sector actually sped up, into the high 20% range, from the mid-20s the previous quarter[3].
There's also a number buried in the filing that neither side likes to lead with. On a GAAP basis — the stricter accounting standard that includes all costs — Datadog's operating income for the quarter was just $5 million, a 0% margin on $1.12 billion of revenue[1]. Strip out one major cost, mostly stock paid to employees, and non-GAAP operating income was $257 million, a 23% margin[1].
That gap, about $252 million, is a real cost. It doesn't leave the company's bank account, but it does shrink existing shareholders' ownership stake, since more shares get issued to cover it. It's a detail that got far less attention on either side of the debate than the customer warning did, even though it shapes how much of Datadog's "profit" is really profit.
The Bulls, The Bears, And The Customer Who Said Nothing
Bank of America kept a Buy rating and a $305 price target, with analysts Koji Ikeda and George McGreehan arguing investors mistook one account's pullback for a company-wide slowdown[9]. Oppenheimer noted Datadog serves more than 750 AI-native customers, including all ten of the biggest AI companies, eight of which spend over $10 million a year[12]. Their case: if a company built into every major AI lab's infrastructure loses volume from just one of those labs, that's a rounding error, not a trend.
The bear case runs the other direction. If losing one customer's usage was enough to cut 7 to 8 percentage points off the growth rate, that customer was doing a lot of the work behind the number everyone had been celebrating[15]. Back in July 2025, Guggenheim downgraded the stock to Sell, warning that OpenAI specifically was building its own monitoring tools and could pull roughly $150 million in revenue away from Datadog[7]. A year later, that warning looks at least partly right. Bears also point to free cash flow margin — the share of revenue left over as cash after running the business — falling from 29% to 25%[15].
The customer at the center of all this hasn't said a word publicly. But the logic of what it's likely doing is straightforward. Running AI systems generates enormous volumes of data, and Datadog charges by volume. At a large enough scale, building basic monitoring in-house can be cheaper than paying a vendor for it indefinitely — the same math that leads big companies to build their own data centers instead of renting cloud space[7][8]. Renewing the contract while trimming usage looks like keeping Datadog for the harder problems and pulling the highest-volume monitoring in-house.
A Rally That Got Mistaken For A Verdict
Here's where the coverage split matters. On August 10, six analyst firms — Citigroup, Scotiabank, Macquarie, Needham, Canaccord Genuity, and Rosenblatt — raised their price targets on Datadog, pushing the average target up about 14%, to roughly $292[11]. The stock rose in response. That's a real move, and a real vote of confidence from Wall Street.
But some outlets reported that Monday bounce as if it were the market's reaction to the earnings themselves — an AI-fueled "breakout," in one trading-education publisher's words, that never mentioned the record-setting crash four days earlier[12]. That's not a small error. It reverses the entire story, turning a partial recovery from history's worst single day for the stock into evidence the market loved the quarter.
Other outlets handled the contradiction more directly. A Hong Kong-based financial site, TradingKey, ran one headline reading "Revenue Up 36% and Guidance Raised, but the Stock Fell 17%" — stating the paradox rather than resolving it[15]. That's arguably the fairest framing available, since resolving it one way or the other means taking a side in a genuine, unsettled argument.
What's Still Unknown
The unresolved question sits with the nine other AI companies Datadog counts as customers, none of whom have said anything publicly[12]. If more of them start moving high-volume monitoring in-house the way Datadog's largest customer apparently has, that caps how far the category can keep growing. If this turns out to be one company's one-time belt-tightening, the effect fades.
Nothing in the public record answers that yet. Datadog itself raised its full-year guidance the same day it disclosed the warning, a signal the company is betting the rest of its business can absorb the hit[1][7]. Whether that bet is right won't be visible in a press release. It'll show up, or not, in the growth rate a quarter or two from now.
Summary
On August 6, 2026, Datadog reported second-quarter revenue of $1.12 billion, up 36% from a year earlier[1]. That beat both the company's own guidance and Wall Street's estimate of about $1.08 billion[4]. Datadog also raised its full-year revenue outlook to between $4.45 billion and $4.47 billion[1]. By the usual measures, it was a strong quarter. The stock then fell roughly 17% to 19% that day — the largest one-day drop in the company's history[4][5][16].
The reason was a single sentence from management. Datadog said its largest customer would reduce usage starting in the third quarter[7][8]. Datadog has never named that customer. Several Wall Street analysts believe it is OpenAI, but that identification is analyst inference, not a company disclosure[7]. The customer had just signed a renewal worth more than $100 million covering 17 Datadog products, and is cutting how much it uses, not leaving[8]. Because Datadog bills mostly by volume — how much data a customer sends it to watch — usage cuts hit revenue directly, even with the contract intact.
Four days later, on Monday, August 10, the stock rose about 10% to 11%[10][11]. That move followed price-target increases from at least six firms, including Citigroup, Scotiabank, Macquarie, Needham, Canaccord Genuity and Rosenblatt, lifting the average target roughly 14% to about $292[11]. It was a partial recovery from the crash, not a reaction to the revenue number. The original framing of this story — that shares jumped about 10% because growth accelerated — merges two separate days with opposite causes.
The genuine dispute is not about the numbers, which nobody contests. It is about what one customer means. Bulls, including Bank of America, argue the market is treating one account's pullback as if the whole business were slowing[9]. Bears argue the episode reveals how much of Datadog's acceleration came from a small number of AI companies, and that those companies can build monitoring tools themselves[7].
The Event
On August 6, 2026, Datadog, Inc. released its second-quarter 2026 financial results and held an earnings call[1]. The company reported revenue of $1.12 billion, up 36% year over year, non-GAAP diluted earnings per share of $0.65, and raised full-year 2026 revenue guidance to $4.45-$4.47 billion[1]. On the call, management said its largest customer would reduce usage beginning in the third quarter, and guided third-quarter revenue to $1.135-$1.145 billion, implying 28% to 29% growth[7][15]. Datadog shares fell roughly 17% to 19% that day, the company's largest single-day decline on record, then rose about 10% to 11% on Monday, August 10, after multiple analyst price-target increases[4][5][16][10][11].
Undisputed Facts
- Datadog reported second-quarter 2026 revenue of $1.12 billion, up 36% year over year[1].
- Non-GAAP diluted earnings per share were $0.65, above the roughly $0.58 analysts expected[4].
- GAAP operating income was $5 million, a 0% GAAP operating margin; non-GAAP operating income was $257 million, a 23% margin[1].
- Datadog raised its full-year 2026 revenue guidance to $4.45 billion to $4.47 billion[1].
- Third-quarter guidance of $1.135 billion to $1.145 billion implies 28% to 29% year-over-year growth, slower than the 36% just reported[15].
- Management said its largest customer, which it did not name, would cut usage starting in the third quarter, after signing a renewal of more than $100 million covering 17 products[7][8].
- Customers paying $100,000 or more per year rose to about 4,720, from about 3,850 a year earlier[1].
- Datadog shares fell about 17% to 19% on August 6, 2026, the largest one-day drop in the company's history, and rose about 10% to 11% on August 10, 2026[4][16][10][11].
The Pressure
Strip away the moralizing and blame. What structural realities persist regardless of which narrative wins?
- Growth stocks are priced on the next year, not the last one
- Datadog's share price reflects expected future revenue, not the revenue just reported. That is why a 36% quarter can be followed by a record decline. What changed was the forecast: third-quarter guidance implies 28% to 29% growth, and the reason given was a specific, named risk[15][7]. The reported number was already assumed.
- Usage-based billing cuts both ways
- Datadog charges largely by volume — the amount of data and number of systems it monitors. During the AI buildout, that meant revenue rose automatically as customers scaled up, with no new sales required. The same mechanism runs in reverse. A customer can stay under contract, keep all 17 products, and still cut its bill sharply[8]. Contract renewals therefore do not protect revenue the way they would under fixed-fee pricing.
- Concentration is invisible until it moves
- Datadog does not disclose what share of revenue its largest customer represents. Investors learned the size of that exposure only when it shrank. Guggenheim had estimated roughly $150 million at risk from OpenAI insourcing back in July 2025 — a specific figure the market largely discounted while the stock nearly doubled[7][4].
- Financial media publishes on price moves, not on causes
- A stock that drops 17% and then rises 10% generates two batches of coverage with opposite emotional tones, often from the same outlets. Aggregators and promotional publishers pick up whichever day fits their frame. That is how a rebound becomes an earnings 'surge.'
Material realityDatadog collected $1.12 billion in the quarter, 36% more than a year earlier, and expects $4.45 billion to $4.47 billion for the full year[1]. About 4,720 customers pay it more than $100,000 a year, up from about 3,850[1]. Those facts do not change with the narrative. Two other facts also persist. First, on a GAAP basis the company earned $5 million of operating income on that $1.12 billion — a 0% margin — because roughly $252 million of costs, mostly stock paid to employees, are excluded from the non-GAAP figure of $257 million[1]. Second, at least one very large customer has decided it can do some of this work itself[7][8]. Whether that decision spreads to the other nine largest AI companies Datadog serves is the single unresolved question, and no public data answers it yet[12].
Narrative as a weaponThree groups are actively shaping how this reads. Datadog wants you to see a platform business with broadening adoption, where one customer's cost-cutting is noise — hence raising full-year guidance on the same day it disclosed the warning. Bullish sell-side analysts want you to see an overreaction and a discount, which is both a real analytical view and the view their firms' business favors; six of them raised targets within four days of the drop. Promotional trading publishers want you to see momentum, and produced the specific distortion at the center of this story — reporting the August 10 rebound as though it were the market's verdict on Q2 revenue, with the record-setting August 6 decline simply absent. The quietest party is the customer itself, which has said nothing and whose identity remains analyst inference. Its silence is why both the bull and bear cases remain arguable on the same set of published facts.
How Each Side Sees It
Each major actor’s view — how it frames things, its underlying incentive, and how it’s materially affected. Tap a side to read it.
Frames it asDatadog's case is that the platform, not any one account, is the business. It says 58% of customers now use four or more of its products, up from 52% a year ago, and 13% use ten or more, up from 7%[3]. Its argument is that once a company routes its monitoring through Datadog, replacing it means rebuilding a lot of plumbing. Management also points to growth outside AI: revenue growth from non-AI customers sped up to the high 20s percent, from the mid-20s the prior quarter and 18% a year earlier[3]. Their strongest claim is that the AI story is a bonus on top of a healthy core, not a substitute for it. The company raised full-year guidance the same day it disclosed the customer warning, which it presents as evidence it believes the rest of the business absorbs the hit[1][7].
WhyDatadog needs to keep a high growth rate to justify a valuation built on future expansion, and it needs employees paid heavily in stock to stay[1]. Disclosing the customer risk early is also a legal and reputational hedge — a surprise later would be far more damaging.
Impact on themThe stock fell to roughly the mid-$230s on August 6 after nearly doubling earlier in 2026[4]. That cuts the value of employee stock compensation and makes acquisitions using shares more expensive. Note the GAAP-versus-non-GAAP gap: $5 million of GAAP operating income against $257 million non-GAAP[1]. Most of that $252 million difference is stock-based pay. It is a real cost to existing shareholders, whose ownership share shrinks — it just does not leave the bank account.
Frames it asTheir argument is that the market made a category error. Bank of America's Koji Ikeda and George McGreehan kept a Buy rating and a $305 target, saying investors are treating one account's slowdown as if the entire company were decelerating[9]. They point out the customer renewed rather than left, and expanded to 17 products[8]. Oppenheimer notes Datadog serves more than 750 AI-native customers and all ten of the largest AI companies, with eight of them spending over $10 million a year[12]. The steelman: if the ten biggest AI firms all run on Datadog, one trimming usage is a rounding item against a category that barely existed three years ago. Six firms raised targets on August 10, pushing the average to about $292[11].
WhySell-side analysts work at firms that earn fees from trading volume and investment banking. A 'buying opportunity' call after a crash is both a genuine view and a business generating one. Long-term holders need a reason not to sell at a loss.
Impact on themThe August 10 rally of about 10% to 11%, closing at $260.78 by one account, recovered a meaningful share of the drop[10][11]. Anyone who bought near the $225 low gained quickly; anyone who sold on August 6 locked in the record loss.
Frames it asThe bear case is about concentration and self-supply. Their central point: if one unnamed customer can knock 7 to 8 percentage points off the growth rate, that customer was a large share of the acceleration everyone was celebrating[15]. In July 2025 Guggenheim downgraded Datadog to Sell, warning that OpenAI was building its own monitoring tools and could remove roughly $150 million of Datadog revenue[7]. That prediction now looks partly borne out. Bears also flag free cash flow margin falling from 29% to 25% — meaning that out of every dollar of revenue, 25 cents is left as cash after running costs and equipment spending, down from 29 cents[15]. Their strongest framing is a question: were you paying a high price for durable growth, or for a handful of AI labs' spending habits?
WhyShort sellers profit directly from declines. Skeptical analysts build reputations on calling tops. Both are structurally motivated to emphasize concentration risk — which does not make the risk imaginary.
Impact on themThe August 6 drop validated the Guggenheim thesis publicly, roughly a year after it was issued and after a period in which the stock nearly doubled and the call looked wrong[4][7].
Frames it asThis party has said nothing publicly about the usage cut, so its position must be inferred with care. The logic available to it is straightforward. Monitoring costs scale with the volume of data a system produces, and AI training and inference produce enormous volumes. For a company running at that scale, building monitoring in-house can be cheaper than paying per gigabyte forever. It is the same calculation that leads large firms to run their own data centers. Renewing a nine-figure contract across 17 products while trimming usage is consistent with keeping Datadog for what it does best and handling the highest-volume pieces internally[8].
WhyAI labs are under heavy pressure to cut the cost of each unit of compute. Vendor bills that scale with usage are a natural target.
Impact on themIf the largest AI companies routinely move high-volume monitoring in-house, that caps the category Datadog has been growing into. If this is one company's cost-cutting in one period, the effect is temporary. Which of those is true is not yet knowable from the public record.
Like this article?
The Bias Ledger average rating 4.7
The same story, as framed by outlets across the spectrum, ordered least to most biased. The bias score (1 = straight, 10 = heavily spun) is an AI assessment of that framing — click an outlet to see its track record. The tell is the word choice or omission that reveals the angle.
| Outlet | Vantage | Bias | How they frame it | The tell |
|---|---|---|---|---|
| Benzinga | U.S. center, retail-investor focused | 3 | 'Datadog Beat Earnings, Raised Guidance — but DDOG Stock Still Plunged 19% as Wall Street Says It All Comes Down to One Customer'[7]. | States the paradox plainly rather than resolving it, which is the fairest available frame. But 'plunged' and the quote-as-headline structure are built for clicks, and 'one customer' flattens a debate about concentration into a single dramatic fact. |
| TheStreet | U.S. center, investor-service focused | 3 | 'Datadog's largest customer renews deal but cuts usage' and, separately, 'Datadog coverage update: BofA stays bullish on weakness'[8][9]. | The first headline is close to a straight statement of the record. The second adopts the analyst's own word, 'weakness,' which quietly frames a 17% decline as a temporary dip rather than a repricing. |
| CNBC | U.S. center to center-left | 4 | Earlier in 2026: 'Datadog stock soars 31% on blockbuster earnings as AI winners emerge in software' — using Datadog as proof the AI trade is real[14]. | The word 'blockbuster' and the framing of a single company's quarter as evidence about an entire investment theme. When the same company's next quarter cut against that theme, it received notably less thematic treatment. |
| TradingKey | Hong Kong-based financial media, non-Western, trading-audience focused | 4 | 'Datadog Q2 2026: Revenue Up 36% and Guidance Raised, but the Stock Fell 17%'[15]; separately 'Revenue Rises 36% as Cash Flow Expands'[3]. | The outlet published two framings within days that point opposite directions — one leading on the contradiction, one leading on cash flow expansion while the free cash flow margin was actually falling from 29% to 25%. Volume-driven publishing rather than an ideological slant. |
| StockStory | U.S., algorithmic financial-content publisher | 5 | 'Datadog Delivers Impressive Q2 CY2026 But Stock Drops 17.3%'[4]. | 'Impressive' is the outlet's own verdict, not attributed to anyone. The construction implies the market got it wrong, without engaging the customer disclosure that caused the drop. |
| Timothy Sykes | U.S., trading-education business; publishes promotional stock content | 9 | 'Datadog Stock Jumps As Analysts Hike AI-Driven Targets' and 'Datadog Stock Jumps As AI Demand Fuels Q2 Breakout,' published August 10[12]. | This is the source of the framing this story was originally assigned. It reports the August 10 rebound of about 9.68% as though it were the reaction to Q2 earnings, and calls it a 'breakout' — never mentioning that the same stock had just posted its largest one-day loss on record four days earlier. Omitting the crash converts a partial recovery into a rally. The publisher sells trading courses, so bullish momentum framing is the product. |
References
- Datadog Announces Second Quarter 2026 Financial Results — Datadog · The company itself — primary source, promotional by nature; leads with record revenue and new products
- Datadog, Inc. Form 8-K, Exhibit 99.1, period ended June 30, 2026 — U.S. Securities and Exchange Commission · Federal regulatory filing archive — primary record, no editorial orientation
- Datadog Inc (DDOG) (Q2 2026) Earnings Call Highlights: Revenue Surges 36% to $1.12B, AI-Driven Growth Accelerates — GuruFocus · U.S. investing-data subscription service; automated earnings-call summaries, bullish default framing
- Datadog (NASDAQ:DDOG) Delivers Impressive Q2 CY2026 But Stock Drops 17.3% — StockStory · U.S. algorithmic stock-analysis publisher, subscription-funded; templated bullish adjectives
- Datadog sinks 21% despite raised guidance, Q2 beat — Yahoo Finance · U.S. mainstream financial portal owned by Apollo Global Management; aggregation-heavy
- Earnings call transcript: Datadog beats Q2 2026 estimates but shares fall 15.6% — Investing.com · Israel-based global financial data and news portal, advertising and brokerage-referral funded
- Datadog Beat Earnings, Raised Guidance — but DDOG Stock Still Plunged 19% as Wall Street Says It All Comes Down to One Customer — Benzinga · U.S. retail-investor financial media, advertising and data-licensing funded
- Datadog's largest customer renews deal but cuts usage — TheStreet · U.S. investor-focused financial media, subscription and advertising funded
- Datadog coverage update: BofA stays bullish on weakness — TheStreet · U.S. investor-focused financial media; this piece relays a sell-side note from a firm with banking relationships
- Datadog Inc Stock (DDOG) Moved Up by 9.99% on Aug 10: Facts Behind the Movement — TradingKey · Hong Kong-based financial media for retail traders; high-volume automated market coverage
- Datadog Shares Surge 11% as Analysts Lift Price Targets — Here are the Key Levels to Watch — AskTraders · UK-based broker-comparison and trading-education site funded by broker referrals; promotional toward active trading
- Datadog Stock Jumps As AI Demand Fuels Q2 Breakout — Timothy Sykes · U.S. trading-education business selling courses and alerts; content is promotional and momentum-oriented
- Datadog tumbles despite Q2 beat; analysts highlight slight deceleration but also see buying opportunity — Seeking Alpha · U.S. crowd-sourced investing platform, subscription-funded; contributor pieces are unedited opinion
- Datadog stock soars 31% on blockbuster earnings as AI winners emerge in software — CNBC · U.S. business network owned by Comcast/NBCUniversal; center to center-left newsroom, market-bullish house tone
- Datadog Q2 2026: Revenue Up 36% and Guidance Raised, but the Stock Fell 17% — TradingKey · Hong Kong-based financial media for retail traders
- DDOG Stock Heads For Record Single-Day Drop Despite Q2 Earnings Beat, Raised Guidance — Stocktwits · U.S. retail-investor social platform and news desk; sentiment-driven, advertising funded