groundcover Series C brings in $100 million for AI observability

groundcover raised a $100 million Series C led by One Peak, bringing the observability startup’s total funding to $160 million as it pushes a customer-controlled approach to AI telemetry. The company, stylized with a lowercase g, said it has more than 250 paying customers and tripled annual recurring revenue over the past year, but it did not disclose valuation, headcount or current ARR.

The round lands in a market dominated by Datadog, Dynatrace, New Relic, Splunk and Grafana, companies with mature products, large integration networks and enterprise support operations. Datadog alone reported more than $3 billion in annual revenue in 2025. groundcover’s pitch is that AI workloads are changing the economics and architecture of observability faster than incumbent platforms were designed to handle.

What is groundcover doing differently in AI observability?

groundcover says enterprises should keep telemetry storage and processing inside their own cloud environments, rather than sending that operational data into a vendor-managed SaaS platform. Its bring-your-own-cloud model keeps the data plane in a customer’s AWS, Microsoft Azure or Google Cloud account, while groundcover runs the managed control plane and product experience. The company also offers a fully self-hosted option.

That setup is tied to pricing. groundcover says it charges mainly by monitored hosts, regardless of telemetry volume, instead of billing by ingestion. CEO Shahar Azulay said during a media briefing that users are limiting, siloing and sampling data because telemetry is growing quickly and existing platforms can become expensive as data volumes rise.

The claim is plausible in dense Kubernetes and AI-heavy environments, where logs, traces, metrics, prompt data, model latency, token use, retrieval pipelines, tool calls and agent behavior can grow faster than infrastructure count. It is not automatically cheaper for every customer. The company’s model is most compelling when telemetry density is high, and less obvious for lightly used fleets spread across many hosts.

Why AI agents are changing telemetry needs

AI-assisted coding and autonomous agents are increasing both the pace of software changes and the amount of operational data enterprises may want to retain. Agents can run multi-step workflows, call tools and interact with production systems, which creates a broader trail of telemetry than conventional application monitoring.

Observability has usually been a post-deployment discipline for human operators investigating incidents through logs, metrics and traces. groundcover argues that observability data will also become feedback for coding agents, giving them production context to spot regressions, assess changes and propose fixes.

The company’s Agent Mode lets engineers query logs, metrics, traces and Kubernetes events in natural language. Azulay said humans still approve production changes, while he expects organizations to grant AI systems more operational autonomy over time.

How eBPF fits into groundcover’s pitch

groundcover uses eBPF, a Linux kernel technology that can observe network traffic, system calls and application behavior with fewer manual code changes. For cloud-native teams, that can reduce the work required to instrument Kubernetes environments and broaden telemetry coverage.

eBPF is not unique to groundcover. Other observability vendors use it as well. groundcover’s argument is that eBPF collection, OpenTelemetry compatibility, customer-controlled storage and host-based pricing are more useful when packaged together.

The company also faces a crowded field. Gartner tracks more than 100 observability products, and major vendors now market AI features for operations, incident response and AI application monitoring. Azulay said groundcover deployments often replace incumbent platforms, but the company has not published independent migration data to validate that claim.

For investors, the bet is that AI systems will make telemetry volume, storage location and billing predictability more important purchase criteria. For customers, the near-term question is narrower: whether keeping observability data in their own cloud and paying by host solves a large enough problem to justify switching from entrenched platforms.

This story draws on original reporting from VentureBeat.