Four Reasons Datadog Is Winning the AI Observability Race

Datadog launched its LLM Observability and AI Integrations suite in late 2024. Within 12 months, these products are contributing an estimated 8% of new annual recurring revenue — a remarkably fast ramp for enterprise software.

The product addresses a genuine pain point. Companies deploying large language models need to monitor token usage, latency, hallucination rates, cost per query, and model performance drift. Traditional APM tools weren't built for this. Datadog built purpose-specific tooling and shipped it before Splunk, New Relic, or Elastic could react.

The Research Desk has covered enough platform shifts to recognise the pattern: the vendor that ships monitoring tooling first for a new infrastructure paradigm captures 60-70% of the long-term market share. We saw this with cloud monitoring, container monitoring, and serverless monitoring. Each time, Datadog won. The AI monitoring race is following the same trajectory.

Datadog's net revenue retention rate recovered to 118% in the latest quarter, up from 112% during the cloud optimisation trough. Customers with $100K+ ARR now number over 3,500, growing 18% year-over-year.

The expansion dynamic is powerful. A typical enterprise starts with infrastructure monitoring, adds APM, then log management, then security, then AI observability. Each product addition increases contract value by 30-50%. Datadog now offers 22 integrated products.

The platform consolidation thesis — enterprises reducing from 5-7 monitoring tools to 1-2 platforms — is playing out. Gartner's latest survey showed 72% of enterprises plan to reduce observability vendors over the next two years. Customers using 6+ Datadog products now represent 48% of ARR, up from 35% two years ago. Each additional product deepens switching costs.

Operating margin reached 25% in 2025, up from 18% in 2023. Free cash flow of $680 million on $2.8 billion revenue represents a 24% FCF margin.

The margin expansion is structural. Each new product shares the underlying data infrastructure, so incremental cost is minimal while incremental revenue is full-price. This is the SaaS platform flywheel at its most effective.

We expect operating margins to reach 28-30% by 2027 as AI observability and security products scale. That puts Datadog in elite company — comparable to ServiceNow and Veeva in margin profile.

At 60x forward earnings, the stock looks expensive in absolute terms. But for a company growing at 25% with expanding margins and a clear path to $5+ billion revenue, the premium is justified by growth durability.

Datadog ingests over 10 trillion data points daily. That volume creates a feedback loop: more data improves anomaly detection, which improves the product, which attracts customers, which generates more data.

The competitors are struggling. Splunk is distracted by Cisco integration. New Relic's consumption pricing transition caused churn. Elastic's open-source model limits monetisation. Dynatrace is the strongest competitor but lacks Datadog's breadth across infrastructure, APM, logs, security, and AI.

Cloud providers' monitoring tools are functional but limited to their own clouds. In a multi-cloud world, Datadog's agnostic platform wins by default.

The AI infrastructure buildout is creating monitoring TAM that didn't exist two years ago, and Datadog is capturing it before incumbents can respond. We've been tracking this market since Datadog's IPO, and the competitive dynamics have never been more favourable.