Datadog, Inc. (NASDAQ: DDOG), the monitoring and security platform for cloud applications, announced the general availability of its GPU Monitoring product on April 22, 2026. This new offering is designed to provide engineering teams with granular visibility into the health and performance of the graphics processing units (GPUs) that power modern artificial intelligence and machine learning workloads. As enterprises increasingly transition AI projects from experimental phases to production environments, the demand for specialized observability tools has grown to address the unique hardware requirements of large language models (LLMs) and generative AI applications.
The GPU Monitoring tool integrates directly into the existing Datadog platform, allowing users to track key metrics such as GPU utilization, memory allocation, temperature, and power consumption. By providing these insights alongside application-level data, Datadog enables organizations to identify bottlenecks in the AI inference and training pipelines. For instance, the product can pinpoint instances where GPUs are idling while still consuming power, or where memory constraints are causing latency in model responses. This level of detail is intended to help companies optimize their infrastructure spend, which has become a significant portion of IT budgets due to the high cost of specialized AI hardware.
Yevgeniy Brikman, Vice President of Product at Datadog, stated that the complexity of the AI stack requires a unified approach to monitoring. He noted that while many organizations have invested heavily in GPU capacity, they often lack the tools to ensure those resources are being used efficiently. The new product aims to bridge the gap between infrastructure management and model performance. The launch includes support for major hardware providers, specifically NVIDIA’s H100 and A100 Tensor Core GPUs, and integrates with container orchestration platforms like Kubernetes and managed services from Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP).
In addition to hardware metrics, the GPU Monitoring solution provides visibility into the software layers of the AI stack. It tracks the performance of drivers, libraries, and frameworks such as CUDA and PyTorch. This holistic view allows developers to correlate hardware performance with specific AI jobs or user requests. Datadog’s announcement also highlighted the inclusion of pre-configured dashboards and automated alerting, which are designed to reduce the time required for teams to set up monitoring for new AI clusters.
The release follows a series of AI-centric updates from Datadog over the past year, including LLM Observability and integrations with vector databases. By adding GPU-level insights, the company now offers a comprehensive suite for monitoring the entire lifecycle of an AI application, from the underlying silicon to the final user interaction. This development comes as industry reports suggest that AI infrastructure spending is expected to continue its upward trajectory, making resource efficiency a primary concern for Chief Information Officers and engineering leaders.