Datadog, Inc. (DDOG)

News Summary

• Agentic AI workloads are driving a measurable observability volume explosion at Datadog: MCP tool calls grew 22x since Q4 2025 and quadrupled quarter-over-quarter, GPU monitoring is being adopted by both AI-native startups and hyperscaler in-house AI labs, and CEO Pomel confirmed that AI inference and agentic CPU workloads are accelerating infrastructure monitoring consumption beyond GPU-attached compute — a direct signal that AI hardware utilization is generating proportionally larger observability demand.
• Datadog's AI customer cohort now exceeds 750, including all top-10 AI leaders, with 31 spending over $1M annually and 8 over $10M; the company landed 7-figure annualized deals with two neuro labs for GPU fleet and model training observability — a use case Pomel called nascent two years ago — and is winning hyperscaler in-house AI labs as new logos, indicating that frontier AI infrastructure operators are standardizing on third-party observability rather than homegrown tooling.
• Bits AI has expanded from alert triage into chat, monitoring management, code generation, synthetic test generation, and release validation, with a new AI credits monetization model rolling out; separately, the Bits Security Analyst agent was decoupled from Datadog's own SIEM to run on third-party SIEMs, targeting a broader AI SOC automation market — and Datadog's second-gen time series foundation model Toto is showing LLM-style scaling laws, with the Adaptive ML acquisition closing in June to accelerate that research.
• Q2 revenue was $1.12B, up 36% YoY, beating consensus by roughly $41M; non-GAAP EPS of $0.65 beat by $0.07; billings grew 38% YoY to $1.18B and RPO rose 43% to $3.47B; free cash flow was $279M at a 25% margin; Q3 guidance of $1.135–1.145B implies a deceleration to 28–29% growth, driven by a deliberate de-risking of a usage reduction from the largest customer, with management stating the business ex-that customer is growing at the same accelerating rate.
• DDOG dropped roughly 17% post-earnings despite the beat and raised full-year guidance to $4.45–4.47B, as the market focused on the Q3 growth deceleration implied by the largest-customer headwind; coverage framed it as a positioning-driven selloff against a stock that had already rallied ~108% YTD, with Zacks maintaining a Buy rating and analysts broadly constructive on the underlying acceleration in non-AI customers (high-20s% YoY, up from 18% a year ago) and enterprise new logo momentum.

Financial Highlights

Earnings Call Highlights

Datadog delivered a blowout Q2 with 36% revenue growth and record sequential dollar adds, driven by broad-based acceleration that management was eager to frame as structural rather than episodic. The dominant tone was confident and expansive, with Olivier Pomel repeatedly emphasizing that AI is a tailwind at every layer of the stack — training, inference, agents, and security — and that non-AI customers are also re-accelerating, now growing in the high 20s year-over-year. Analysts were largely constructive but probed hard on the largest customer usage reduction, which management fully derisked in guidance while insisting the core business is booming. The call had a notable agentic AI thread running through it, with MCP tool calls growing 22x since Q4 2025, signaling that the observability surface area is expanding rapidly into new AI workloads. Management posture was bullish and forward-leaning, with Bits AI, GPU monitoring, and agent observability positioned as the next major growth vectors.

• MCP tool calls quadrupled quarter-over-quarter and grew more than 22x compared to Q4 2025, and the number of AI customers using Datadog now exceeds 750, including all 10 of the top 10 AI leaders — with 31 AI customers spending more than $1 million annually and 8 spending more than $10 million annually.
• Datadog landed 7-figure annualized deals with two neuro labs specifically for GPU fleet and model training observability, a use case Pomel described as 'not really a business area for us a couple of years ago,' and separately disclosed that hyperscaler in-house AI labs are now also being won as new logos.
• Bits AI has been decoupled from its original alert-investigation use case and now spans chat, monitoring management, code generation, synthetic test generation, and release validation; Datadog is rolling out an AI credits packaging model to monetize the expanded surface area, and Pomel noted that Bits AI usage drives more product deployment rather than cannibalizing it.
• The Bits Security Analyst agent has been separated from Datadog's own SIEM so it can run on third-party SIEMs, a deliberate move to pursue a broader AI SOC automation market rather than limiting the agent to customers willing to re-platform their SIEM.
• Datadog's second-generation time series foundation model Toto demonstrated scalability analogous to LLM scaling laws, and the company is accelerating research into larger dedicated models and multi-modal world models, supported by the acquisition of Adaptive ML closing in June.

Sources

• Datadog (DDOG) Q2 Earnings and Revenues Surpass ...
• Datadog Announces Second Quarter 2026 Financial Results
• Datadog (DDOG) Q2 Earnings: Taking a Look at Key ...