groundcover Review (2026): Pricing, Features & Honest Verdict

TLDR

groundcover is an eBPF-based observability platform for Kubernetes that bills per node instead of per gigabyte, and stores your telemetry inside your own cloud account. It is the most credible Datadog escape route we have looked at. Best for: engineering teams on Kubernetes with a runaway observability bill. Price: Free plan available, Pro $30/node/mo. Rating: 7.6/10

What is groundcover?

groundcover is a full-stack observability platform: logs, metrics, traces, events, real user monitoring and now AI workload monitoring, in one place. The difference is how it collects and how it charges. Collection happens through an eBPF sensor that reads traffic at the Linux kernel level, so you get application-level visibility without adding SDKs, agents or code changes to your services. Billing happens per Kubernetes node per month, not per gigabyte ingested, so sending more data does not cost more money.

The company was founded in 2021 and is based in Tel Aviv. In 2026 it raised $100 million at a $500 million valuation with an explicitly stated goal of replacing Datadog. The architecture backs that up: groundcover is BYOC by default, meaning the data plane runs inside your own AWS or GCP VPC and your telemetry never leaves your account. We dug into it because observability is the line item that quietly becomes the third biggest bill in most engineering orgs, and groundcover is one of the few vendors attacking the pricing model rather than just the feature list.

What Are groundcover’s Key Features?

Three of these are genuinely differentiated and one is table stakes with a good implementation. Here is what stood out.

Zero-Instrumentation eBPF Sensor

You install a DaemonSet and groundcover starts seeing HTTP calls, database queries, gRPC traffic and payloads across every pod on the node. No OpenTelemetry SDK to wire in, no redeploys, no per-language agent. This is the fastest time-to-value of anything in the category: teams report useful traces within minutes rather than a sprint. The requirement is Linux kernel 4.18 or newer on your nodes. If you are running older kernels you have an upgrade to do first. Coverage is broad but not total: eBPF support for MSSQL is missing, and navigating deep distributed traces is one of the rougher parts of the UI.

BYOC Architecture

The data plane, the storage and optionally the UI all run inside your own cloud account. Nothing gets shipped to a vendor bucket. For anyone who has failed a security review because logs contained PII, or who works under data residency rules, this removes the whole argument. The trade-off is that you now operate that infrastructure and pay your own cloud provider for its storage and compute. groundcover manages the control plane, but the bytes are your problem. That is a genuine operational commitment, and teams new to BYOC should expect a learning curve on capacity planning for the observability stack itself.

Node-Based Pricing With No Sampling

This is the commercial core of the product. groundcover charges a flat rate per actively monitored Kubernetes node per month, calculated on a monthly average, and any node size counts the same. Because you are not paying by volume, there is no incentive to sample traces, drop debug logs or set aggressive retention. Real user monitoring is folded into the node license rather than billed per session, which is where competitors quietly extract a lot of money. Everything queries through one language across logs, traces, metrics and events, so you are not learning three query syntaxes to answer one question.

Agent Mode

groundcover ships an AI agent that investigates incidents, builds dashboards and creates monitors from natural language prompts. Ask it why latency spiked on a service and it walks the correlated signals rather than making you pivot between four tabs. It is genuinely useful and it is also the feature most likely to be matched by every competitor within a year. Do not buy on this alone, but it is a nice reason to prefer groundcover over an open source stack you assemble yourself.

How Much Does groundcover Cost?

groundcover starts free and paid plans begin at $30 per host per month. The Free tier is $0, requires no credit card, gives you 12-hour data retention and Slack community support, which makes it a real evaluation environment rather than a marketing gate. Pro is $30 per host per month and adds all integrations and data sources, SSO and standard retention. Enterprise is $35 per host per month and adds RBAC, unlimited data retention and premium support. On Premise is $50 per host per month for a fully self-hosted data plane, self-hosted UI and isolated authentication. Billing is based on the monthly average of Kubernetes nodes actively monitored by the eBPF sensor.

Compare that to Datadog, where Infrastructure Pro is $15 per host per month and Enterprise is $23, which looks cheaper until you add the rest. Container monitoring is $1 per container per month, logs are billed per gigabyte ingested and again per gigabyte indexed, APM is a separate line, RUM is billed per session, and Bits AI SRE investigations run $500 per month for 20 investigations. The Datadog headline number is never the bill. groundcover’s $30 covers logs, metrics, traces, RUM and events with no volume component, so a chatty cluster and a quiet cluster cost the same.

The catch is that node-based pricing punishes the opposite shape of workload. If you run many small nodes with low traffic, you pay for node count you are not using. If you run few large nodes pushing huge telemetry volume, groundcover is dramatically cheaper. Independent 2026 comparisons put the effective rate at roughly $25 to $40 per node per month depending on contract length. Do the arithmetic against your actual node count before you assume it is a saving. One more thing to factor in: because groundcover stores data in your account, the cloud storage cost sits on your own bill rather than the vendor invoice. It is usually small relative to what you save, but it is not zero, and it grows with the unlimited retention that Enterprise unlocks.

Who is groundcover Best For?

Use groundcover if you run production workloads on Kubernetes and your Datadog or New Relic bill has become a recurring topic in finance meetings. It is also the right pick if you have ever been told to sample traces or cut log retention to control cost, because the pricing model removes that pressure entirely. Teams with data residency requirements, or who cannot ship telemetry to a third-party cloud, will find BYOC hard to beat. The free tier makes the evaluation cheap: install the DaemonSet on a staging cluster, watch traces appear without touching application code, and you will know within an hour whether the eBPF approach covers your stack.

Skip groundcover if:

• You are not on Kubernetes. The product is built around it and the value drops sharply on plain VMs or serverless-only stacks.
• You need continuous profiling with CPU flame graphs per service. Datadog and Dynatrace have it, groundcover does not.
• You rely on pushing arbitrary custom business metrics from application code. You cannot do that the way you can with StatsD or Prometheus, so you will still run Prometheus alongside.
• You depend on a long tail of niche integrations. After 15 years Datadog has hundreds; groundcover, founded in 2021, does not.
• Your nodes run kernels older than 4.18, or you cannot upgrade them.

Best groundcover Alternatives

Datadog

Datadog is the incumbent and the most complete platform in the category. Infrastructure Pro is $15 per host per month, Enterprise $23, with container monitoring at $1 per container per month and logs, APM and RUM billed separately on top. Pick Datadog when you need breadth of integrations, continuous profiling, or a vendor your enterprise customers already trust. Pick groundcover when you have run the numbers and the per-gigabyte components are what is actually killing you.

Grafana Cloud

Grafana Cloud has the most generous free tier in observability: 10,000 active metric series, plus 50 GB each of logs, traces and profiles per month. Pro adds a $19 per month platform fee plus usage, at $6.50 per 1,000 billable series and $0.40 per gigabyte to write logs. It is the flexible, open-standards choice and it does have profiling. The downside is that you assemble it yourself and the usage-based bill has the same unpredictability groundcover is trying to escape.

SigNoz

SigNoz is the open source option. The Community Edition is free to self-host with no data cap, and SigNoz Cloud starts at $49 per month including $49 of usage, roughly 163 GB of logs and traces or 490 million metric samples, at $0.30 per gigabyte thereafter. It is OpenTelemetry-native, which means you instrument your code rather than getting it free from eBPF. Choose SigNoz if you want no vendor lock-in and have the engineering time; choose groundcover if you would rather not spend that time.

Final Verdict: Is groundcover Worth It?

groundcover scores 7.6 because it is the rare product where the pricing model is the feature. eBPF collection means you get deep visibility without instrumenting anything, and node-based billing means nobody on your team ever has to argue about sampling rates again. If your observability spend has been growing faster than your traffic, this is the first vendor to put on the shortlist. The free tier with no credit card makes evaluating it almost free of risk.

It is not a clean sweep. The product is Kubernetes-first to the point of being Kubernetes-only in practice, there is no continuous profiling, custom metrics are limited enough that most teams keep Prometheus running alongside, and the integration catalog is thin next to a vendor that has been at it since 2010. BYOC also means you inherit real infrastructure responsibility. Our recommendation: if you run Kubernetes at any meaningful scale, install the free tier this week and compare the projected node-based cost against your current bill. If the saving is large, the missing features are a fair trade. If you are not on Kubernetes, look elsewhere.