How AI is Transforming Enterprise Observability: The Rise of groundcover
Enterprise AI Telemetry and Cloud Observability: groundcover’s $100M Raise
Observability startup groundcover announced this week that it raised $100 million in a round led by One Peak, bringing its total funding to $160 million. As enterprise AI adoption scales in production environments, the company reports holding more than 250 paying customers and tripling annual recurring revenue over the past year while attempting to disrupt traditional monitoring platforms like Datadog, Dynatrace, New Relic, Splunk, and Grafana.
The Tech TL;DR:
- Funding Milestone: groundcover secured a $100 million round led by One Peak, pushing its total funding to $160 million.
- Architectural Shift: The platform utilizes a bring-your-own-cloud (BYOC) model and eBPF kernel sensors to keep telemetry data storage within customer environments.
- Pricing Dynamics: Pricing is based on monitored host capacity rather than data ingestion volume, directly targeting escalating enterprise observability bills driven by AI workloads.
Shifting Telemetry Architecture for AI Workloads
Artificial intelligence has fundamentally altered the assumptions underlying modern software observability. Historically viewed as a post-production discipline where engineers monitor logs, metrics, and traces to investigate incidents, observability now faces an influx of operational data generated by automated coding assistants, complex distributed microservices, Kubernetes clusters, and autonomous AI agents.
According to groundcover co-founder and CEO Shahar Azulay, telemetry is exploding across enterprise systems. Speaking during a recent media briefing, Azulay noted that users are increasingly frustrated by limitations in established platforms that force them to silo, sample, or truncate operational data to manage costs. Traditional pricing models charge organizations according to the volume of data ingested, creating a direct conflict with AI applications that generate massive streams of token usage, model latency data, prompt executions, and tool invocations.
BYOC Architecture and eBPF Implementation Mechanics
To bypass ingestion-based pricing models, groundcover employs a bring-your-own-cloud (BYOC) architecture alongside a fully self-hosted deployment option. Customers maintain their data plane—including telemetry storage and processing—inside their own Microsoft Azure, AWS, or Google Cloud environments, while the vendor supplies a managed control plane. This contrasts with traditional SaaS observability stacks where telemetry is processed and stored on vendor-managed infrastructure.
By keeping the data plane inside the enterprise perimeter, groundcover prices its software based on monitored host count rather than gigabytes or terabytes ingested. This model changes how infrastructure teams handle data retention, allowing organizations to retain complete operational telemetry without unpredictable cost fluctuations. Under the hood, the platform relies on extended Berkeley Packet Filtering (eBPF), a Linux kernel technology that observes network traffic and application system calls directly at the OS level.
By capturing metrics via eBPF kernel hooks rather than manual application code instrumentation, engineering teams can deploy monitoring infrastructure across Kubernetes clusters with minimal overhead. Organizations migrating complex workloads often coordinate these deployments alongside specialized software development agencies to ensure seamless containerization and cluster stability.
Agent Mode and the Shift Toward Autonomous Operations
As enterprises begin experimenting with autonomous systems, observability is shifting from a tool used exclusively by human operators to an operational feedback loop for AI agents. groundcover’s Agent Mode enables engineers to investigate production incidents using natural language across metrics, logs, traces, and Kubernetes events.
According to Azulay, production telemetry is increasingly fed back into coding agents so autonomous systems can evaluate changes, identify regressions, and recommend fixes. While human operators currently remain in the loop to approve production changes, the long-term architectural goal is to provide intelligent software with the real-time operational context it needs to manage distributed systems autonomously.
Market Competition and Deployment Realities
Breaking into the enterprise observability market remains a steep challenge. Established incumbents maintain billions of dollars in annual revenue, extensive partner ecosystems, and decades of product maturity. Datadog alone generated more than $3 billion in annual revenue in 2025, while Dynatrace, Cisco’s Splunk business, Grafana Labs, and New Relic maintain deeply entrenched enterprise positions.
Industry analysts note that groundcover’s reported growth metrics and customer adoption figures originate from the company itself, and vendor-authored case studies reporting significant cost savings require independent validation. Furthermore, enterprise IT departments evaluating BYOC architectures must scrutinize exactly what metadata leaves their cloud environments during standard operations. Organizations reviewing their compliance postures frequently engage vetted cybersecurity auditors and penetration testers to review data egress policies and ensure strict adherence to internal security standards.