Datadog says AI is widening its market as enterprise cloud demand accelerates

Pomel frames the AI shift as an infrastructure expansion

Datadog Chief Executive Officer Olivier Pomel said the company grew roughly 36% year over year in its most recent quarter, extending a pattern of sequential acceleration. Speaking at the Citi enterprise software conference, Pomel argued that the current move toward artificial intelligence resembles the earlier migration to cloud computing, forcing companies to operate more digital infrastructure, ship software faster and manage increasingly interconnected development, operations and security functions. He estimated that 70% to 80% of the technology stack used by AI companies resembles infrastructure used by non-AI companies, but said AI deployments add new layers to monitor, including GPUs, models, agents, agent outcomes, security and interactions with external systems. "The part of the business growing the fastest is the part that has to do with the buildup at the infrastructure and the application layer," Pomel said.

Non-AI customer growth accelerates to the high 20% range

The expansion is not limited to AI-native businesses. Pomel said Datadog's non-AI customer segment accelerated to growth in the high 20% range, up from about 18% a year earlier. Datadog holds roughly 13% to 14% of the observability market, which Pomel described as expanding at annual rates ranging from 15% to 30% depending on the source. The company serves about half of the Fortune 500 and is working to deepen those relationships. Datadog continues to invest about 30% of revenue, or more than $1 billion, in research and development as it competes with rivals including Dynatrace.

Obstler outlines a predictive, eventually automated, observability platform

Chief Financial Officer David Obstler, speaking at a company event, said Datadog is investing in proprietary models, open-source tools and foundation models to make observability more predictive. The company recently acquired Adaptive ML, a specialist in reinforcement learning for IT management and observability, and has established a research lab. Datadog previously released a model called Toto, with a newer version, Toto 2.0, also available. Obstler described a long-term vision in which Datadog can generate more accurate real-time signals, improve application functionality and potentially support automated remediation, with customers deciding how much authority to delegate to the platform. "That has tremendous ramifications, both in terms of the speed and also the efficiency in human capital," Obstler said. He said the company is not yet at that point, but its Bits product line is intended to use models and data to become more predictive.

AI-native workloads and enterprise retention

Obstler characterized AI-native companies as cloud-native businesses investing in modern applications, operating without legacy infrastructure and experiencing accelerated demand, and said Datadog is seeing strong workload growth from those customers as well as a movement from internally built or pieced-together observability systems toward Datadog's platform. He said thousands of customers are using the company's Agent Observability offering and that related revenue streams are beginning to emerge. While Datadog initially expected its primary AI opportunity to center on production inference workloads, it has also received requests from larger hyperscalers and other companies to support training and post-training workloads. "The speed of this is such that the pre-production or late training into production is starting to merge," Obstler said, adding that the company does not have evidence to conclude it will become "the training company." On customer retention, Obstler said Datadog's gross retention has historically been in the upper 90% range, with larger enterprises at the high end. He said the company's largest customer renewed and extended its commitment contract, and that Datadog has incorporated only the commitment level from that customer into its guidance because expansion and consumption can vary.