Datadog went into its August 6, 2026 earnings report having nearly doubled in price over the prior twelve months. It came out having lost almost a fifth of its value in a single session. The quarter was, by every traditional measure, excellent. Wall Street’s reaction was not irrational. It was a very precise calculation about one number that did not appear in the headline results.
The Big Question
Can a cloud software company that monetizes AI workloads be penalized for being too dependent on AI customers? Datadog’s Q2 answer is yes, and the mechanism is subtler than most post-earnings commentary has captured. The question now sitting in every investment committee that owns the stock: is this a one-quarter air pocket or the first visible crack in the consumption-based model that powered DDOG’s extraordinary run?
Why Wall Street Cares
Datadog reported adjusted earnings of $0.65 per share against an estimate of $0.58, and revenue of $1.12 billion against a forecast of $1.08 billion. Revenue growth accelerated to 36% year over year, and sequential revenue growth of 11% was the strongest since Q2 2022, according to management. The company also raised full-year guidance for the third consecutive quarter. On any ordinary day, that is a clean beat-and-raise.
Despite the 36% revenue surge and the guidance raise, the stock plunged 19% on August 6 as weak third-quarter guidance tied to a single major client sparked investor panic. The reason sits inside a single line from the earnings call that most headline readers missed.
The Bull Case
The platform breadth argument is compelling and the data supports it. Fifty-eight percent of customers now use four or more Datadog products, up from 52% a year ago. Non-AI customer revenue growth accelerated to the high 20s percent year over year, up from the mid-20s last quarter and 18% a year ago. Enterprise new-logo bookings more than doubled from a year ago, and newer customers are ramping faster: about 30% of year-over-year growth now comes from customers landed in the past year, up from 25% last quarter.
Billings grew 38% year over year to $1.18 billion, and remaining performance obligations rose 43% year over year to $3.47 billion. Those two figures tell the forward revenue story more honestly than any single quarter’s recognized revenue. Obligations at $3.47 billion growing 43% means the pipeline is accelerating, not slowing.
The product argument extends further. Of Datadog’s 26 products, five generate over $100 million in ARR each, three more are between $50 million and $100 million, and 18 are still early in their lifecycle. That is a platform business with a long tail of monetization still ahead of it.
The Bear Case
The problem is structural, not cyclical, and it lives inside how Datadog charges its largest customers. A customer can renew a multi-year agreement but generate less near-term revenue if it monitors fewer servers, sends fewer software records, or consumes less capacity than Datadog expected. The nine-figure renewal protects the customer relationship, but it does not protect all revenue linked to actual consumption or usage above the contracted amount.
Datadog’s largest customer, a leading AI company with a nine-figure contract using 17 Datadog products, reduced its usage starting in the third quarter, despite a recent renewal. While CEO Olivier Pomel refused to explicitly name the client on the earnings call, the leading AI company is widely believed by Wall Street analysts to be OpenAI. The year-ago warning is worth revisiting here. In July 2025, Guggenheim downgraded the stock to sell, warning that OpenAI was developing in-house monitoring tools that could carve a $150 million hole in Datadog’s revenue. The stock ignored that call and kept climbing. The stock brushed it off and nearly doubled over the following year. On August 6, the warning arrived anyway, delivered not by a bearish analyst but by Datadog’s own management.
The detail that moved the stock was Datadog’s third-quarter revenue guidance: $1.135 billion to $1.145 billion, which implies year-over-year growth of 28% to 29%, compared to the 36% growth the company had just reported for Q2. Seven to eight percentage points of deceleration in a single quarter, for a stock priced at a significant premium to growth, is exactly the kind of number that triggers institutional repositioning.
The Evidence
The earnings call transcript is where the real investment debate lives. CEO Pomel confirmed the nine-figure renewal with a leading AI company, noting the customer uses 17 Datadog products to enable unified visibility on production workloads at very large scale, albeit with a usage reduction starting in Q3. That phrase, usage reduction, is doing significant work. Management noted that customers, especially CFOs, are increasingly focused on software value and cost control.
Gross margin declined to 79.6% in Q2, down from 80.2% in the previous quarter and 80.9% in the year-ago quarter. That compression is modest but directionally problematic for a company whose premium multiple depends on expanding margins over time. Companies that miss Wall Street’s estimates for earnings per share and sales have seen an average one-day decline of 7.4% this earnings season, per Evercore ISI, far exceeding the typical drop over the past five years of 3.2%. DDOG beat both and still fell more than twice the miss penalty. That says something specific about where institutional holders were positioned coming into August 6.
The CFO’s language on the call added another layer. In addressing the largest customer, management said, “We have seen a usage reduction which is incorporated in our Q3 and full year 2026 guidance.” CEO Pomel noted that the company “fully de-risked the guidance for the rest of the year” following the usage reduction. De-risking is reassuring language. It also means the Q3 guide already has the damage baked in. If usage falls further, there is no cushion.
The Mavens’ View
The Street is split in a way that is itself instructive. Wall Street firms including Citi, Morgan Stanley, Baird, BMO, Raymond James, Canaccord, Needham, and Cantor Fitzgerald raised DDOG price targets into the $280 to $327 band while shares traded in the mid-$230s. That is a wide gap between target and price, which reflects genuine disagreement about whether the selloff was a pricing error or a signal.
Bank of America stayed bullish and called the decline a buying opportunity. Futurum Equities’ Shay Boloor observed that the massive software sector selloff “comes down to one customer,” highlighting the inherent risks of Datadog’s consumption-based business model where large enterprises treat cloud and AI workloads like a utility bill. That framing is useful. If you believe the one-customer problem is a one-quarter adjustment, the stock looks cheap against its RPO. If you believe it reflects a broader AI-native efficiency drive, the multiple compression has further to run.
Pomel noted that excluding the largest customer, the business is growing at the same rate and accelerating steadily. That is the clean version of the bull case. Analysts highlight record ARR additions, more than double new-logo bookings, and strong non-AI revenue growth in the mid-20% range, signaling broad demand beyond the AI hype. The bear case is not that the core business is broken. It is that the core business, absent the one super-customer, cannot justify the multiple the stock carried into August 6.
What Investors Are Missing
The consumption-based model is a double-edged surface that the market has not fully priced in. When AI workloads explode, Datadog’s revenue scales with them automatically, no sales motion required. When those workloads optimize, the revenue contracts just as automatically, also with no warning. Datadog did not reveal which billing arrangement applies to its largest customer. Investors therefore cannot determine how much of the contract is recognized evenly and how much depends on usage. That opacity is not fraud. It is the structural information asymmetry baked into usage-based pricing at scale.
The deeper question is whether the AI-native cohort is fundamentally different from the enterprise cohort. The AI-native cohort, which includes Datadog’s largest customer, grew only high single digits year over year in Q1 2026. That was already a sign. The real AI beneficiary thesis is that AI is making every company ship more code, deploy more apps, and generate more telemetry. Traditional enterprises are building agents, spinning up GPU clusters, and running inference. Every one of those workloads needs to be monitored. That thesis holds if the enterprise cohort absorbs the AI-native deceleration. The next two quarters decide whether it does.
There is also an MCP signal buried in the release that most analysts glossed over. Management said MCP tool calls increased fourfold quarter over quarter and more than 22x compared to Q4 2025. That is agentic activity inside Datadog’s own platform growing at a pace that is difficult to model. It could be the next consumption driver. It could also be monetized at a different unit price than traditional observability seats, with margin implications nobody has yet quantified.
Stocks to Watch
Datadog (DDOG). The obvious name. The stock fell almost 17%, closing near $234 in its steepest single-day drop in years. By August 10, it had clawed back a little over half the dollars lost, closing at $260.78, up 11.48%, though still below its pre-earnings level. The recovery argues the selloff was overdone. The unresolved usage question argues it may not be over.
Dynatrace (DT). The most direct competitive read on whether Datadog’s one-customer problem is company-specific or sector-wide. If Dynatrace’s enterprise accounts show the same CFO cost-scrutiny pattern when it reports, the observability sector re-rating accelerates. If they do not, DDOG’s problem stays idiosyncratic.
Cisco (CSCO). Cisco acquired Splunk and is now the incumbent bundling observability, networking, and security into enterprise contracts. Splunk is now inside Cisco and pushing aggressive bundling on logs and security. Every dollar of enterprise budget that Cisco captures through bundling is a dollar Datadog’s land-and-expand motion cannot reach. The two companies reported the same week. Cisco beat estimates and also fell. The reasons were different. The overlap in enterprise buyers is not.
Elastic (ESTC). The open-source alternative that AI-native companies evaluating cost reduction will compare against Datadog’s per-seat, per-product pricing. If OpenAI’s usage reduction is partly a migration toward open-source tooling, Elastic is the passive beneficiary of the same math that hurt DDOG.
The next few quarters will show whether Datadog’s other 4,700-plus large customers can grow fast enough to make one account’s habits irrelevant to the stock. If they can, this selloff becomes the buying opportunity BofA is calling it. If they cannot, Guggenheim’s year-old downgrade will look less like a bad call and more like an early one. That is the precise framing every portfolio manager should be using right now.
