Grafana’s cost-efficiency scales against Datadog and New Relic

The cost of scaling telemetry

Grafana Labs surpassed $400 million in annual recurring revenue by the end of the fiscal year ending January 31, 2026. The company supports over 7,000 organizations, which include 70% of the Fortune 50. As telemetry volumes grow five to ten times higher due to service proliferation, SRE teams face massive budget risks from traditional pricing models. Datadog charges separately for hosts, custom metrics, log ingestion, and APM traces, and its high-water mark infrastructure billing can lead to unpredictable costs. New Relic uses a consumption-based model that avoids per-host charges but can produce surprising bills at very high scales. Grafana Labs addresses these financial pressures with its Adaptive Telemetry suite to manage spend across metrics, logs, traces, and profiles. The 2026 evaluation cycle focuses on whether a platform can answer questions about systems not yet built, especially as AI workloads on Kubernetes produce telemetry like token consumption and GPU utilization per inference task. Grafana also utilizes Loki and Tempo to provide a composable architecture of separately focused components. Loki indexes metadata of logs rather than the lines themselves, while Tempo uses an object-storage-backed retention model to reduce costs. The company continues to grow its global presence, and it added 100 employees and a new subsidiary in Japan to accelerate market growth.

Comparing feature sets and AI

Grafana 12 provides observability as code through the Git Sync feature, which mirrors dashboard and folder configurations to a Git repository. This functionality contrasts with the broad integration ecosystem of Datadog, which maintains more than 1,000 integrations. While Datadog provides the Bits AI SRE agent for $500 annually to perform 20 investigations per month, Grafana utilizes the Grafana Assistant to help users generate dashboards and queries. Mimir 3.0 also debuted to deliver improvements in performance and scalability for large-scale metrics workloads, while Beyla provides eBPF-based auto-instrumentation at the network level. Datadog includes Watchdog for monitoring and Universal Service Monitoring via eBPF for service discovery without code changes. Grafana Cloud provides a free tier with up to 50 GB of logs, while New Relic offers a 100 GB per month free ingest tier.

Datadog manages high-cardinality Kubernetes labels through specific billing decisions, but custom metrics beyond the 100 per host allotment cost $0.05 per metric per month. Datadog also expanded LLM Observability with integration with Amazon Bedrock. Dynatrace competes through its Davis AI engine, which catches anomalies through causation-based root cause analysis. Dynatrace remains the most expensive commercial option at scale, and its monolithic agent model creates friction for teams that want fine-grained control over instrumentation.

The move to open standards

The industry moves toward OpenTelemetry to avoid instrumentation lock-in, and Grafana remains a leader in this movement because the company contributes heavily to the OpenTelemetry and Prometheus projects to maintain interoperability. Companies like Anthropic and NVIDIA rely on Grafana Cloud to monitor AI workloads, specifically tracking LLM agent activity and model inference. You should prioritize query performance and cost behavior at scale when evaluating these platforms. Datadog provides high-quality granularity in its infrastructure monitoring, yet its complex pricing structure for hosts and custom metrics creates significant budget unpredictability for teams managing large production environments at scale. New Relic remains the price-performance winner for mid-size teams because its data ingest pricing stays transparent. Its Browser Monitoring tool captures the four golden signals of user experience. The New Relic Database stores all signal types in a unified telemetry database, and New Relic provides core functionality out-of-the-box, although it lacks advanced features like session replay that Datadog offers. Can SRE teams effectively manage the per-series billing risk of $6.50 per 1,000 active series in Grafana?