OpenTelemetry Cost Guide: Is Open-Source Observability Really Free?

The OTel Promise vs Reality

OpenTelemetry is a CNCF graduated project with 10,000+ contributors from 1,200+ companies. It's the industry standard for instrumenting applications and collecting telemetry data — see the OpenTelemetry project site for specifications and SDKs. And yes — the framework itself is 100% free.

But "free" is misleading. The real question isn't whether OpenTelemetry costs money to use. It's: what does a production-quality observability stack built on OpenTelemetry actually cost to run?

The answer depends on three factors:

1. Which backend you send data to
2. How much infrastructure the Collector requires
3. How much engineering time you invest

What You're Actually Paying For

OpenTelemetry Framework: $0

The instrumentation libraries, the Collector, the protocol (OTLP) — all free and open source. No per-host fees, no per-user licenses, no usage limits.

Backend Options: $0 to $25,000+/month

This is where 90% of your observability cost lives. Your options:

Backend	Type	Monthly Cost (Mid-size)	Pros	Cons
Jaeger + Prometheus + Loki	Self-hosted OSS	$300-800 + eng time	Full control, cheapest	Operational overhead
Grafana Cloud	Managed OSS	$500-2,000	Easy, generous free tier	Less customizable
SigNoz Cloud	Managed OTel-native	$400-1,500	Built for OTel, simple pricing	Smaller ecosystem
New Relic	Commercial	$500-5,000	100GB free, good UX	Per-user pricing adds up
Datadog	Commercial	$2,000-15,000	Best integrations	Most expensive

Collector Infrastructure: $50-$2,000/month

The OTel Collector needs to run somewhere. At small scale, it's a sidecar or a single gateway. At enterprise scale, you need a dedicated Collector cluster.

Rough infrastructure costs for the Collector itself:

Scale	Setup	Monthly Cost
Small (10 services)	Single gateway instance	$50-100
Medium (50 services)	3-node Collector cluster	$200-500
Large (200+ services)	Dedicated Collector fleet	$500-2,000

Engineering Time: The Hidden Multiplier

This is what most OTel cost analyses ignore. Someone needs to:

• Configure and maintain the Collector pipeline
• Troubleshoot data gaps and sampling issues
• Manage upgrades (OTel releases frequently)
• Tune sampling, filtering, and routing rules
• Train the team on new query languages

Estimate: 0.25 to 1.0 FTE depending on scale and complexity. At $150K/year fully loaded, that's $37K-$150K/year in engineering overhead.

Total Cost of Ownership by Scale

Small Team (5 engineers, 10 services)

Component	Self-Hosted	Managed (Grafana Cloud)	Datadog
Backend	$150/mo	$45/mo (free tiers)	$1,150/mo
Collector infra	$50/mo	$50/mo	$0 (bundled)
Engineering (0.1 FTE)	$1,250/mo	$625/mo	$0
Monthly Total	$1,450	$720	$1,150
Annual Total	$17,400	$8,640	$13,800

At small scale, managed OTel backend wins. Self-hosting doesn't save enough to justify the engineering time. Datadog is competitive because the operational overhead is zero.

Mid-Size (30 engineers, 50 services)

Component	Self-Hosted	Managed (Grafana Cloud)	Datadog
Backend	$800/mo	$1,455/mo	$8,350/mo
Collector infra	$300/mo	$300/mo	$0
Engineering (0.5 FTE)	$6,250/mo	$3,125/mo	$0
Monthly Total	$7,350	$4,880	$8,350
Annual Total	$88,200	$58,560	$100,200

At mid-size, managed OTel backend is the sweet spot. Self-hosted is cheaper on vendor costs but the engineering time tips the balance. Datadog is 40-70% more expensive.

Enterprise (150 engineers, 200+ services)

Component	Self-Hosted	Managed (Grafana Cloud)	Datadog
Backend	$2,500/mo	$11,615/mo	$54,900/mo
Collector infra	$1,500/mo	$1,500/mo	$0
Engineering (1.0 FTE)	$12,500/mo	$6,250/mo	$0
Monthly Total	$16,500	$19,365	$54,900
Annual Total	$198,000	$232,380	$658,800

At enterprise scale, self-hosted OTel wins on total cost — if you have the platform engineering team to run it. The $460K annual savings vs Datadog easily funds a 2-person observability platform team.

The Strategic Play: Instrument First, Optimize Later

Regardless of your current backend, instrumenting with OpenTelemetry today is the right move:

1. You decouple instrumentation from your backend choice
2. You can switch backends without touching application code
3. You gain access to pipeline-level cost optimization (sampling, filtering, routing)
4. You future-proof against vendor price increases

The migration path:

1. Month 1: Deploy OTel instrumentation alongside existing agents
2. Month 2-3: Validate data quality and completeness
3. Month 4-6: Migrate to OTel-only and remove vendor agents
4. Month 6+: Optimize backend choice based on real cost data

Bottom Line

OpenTelemetry isn't free — but it gives you control over where your money goes. The framework itself costs nothing. The real cost is your backend choice and engineering investment.

For most teams, the answer is: instrument with OTel, send to a managed backend, and self-host only when the savings justify the operational investment.

Sources & further reading

• OpenTelemetry documentation — instrumentation, Collector, and OTLP protocol
• CNCF: OpenTelemetry — project governance and adoption status
• Elastic Observability pricing — managed Elastic Cloud rates for comparison

Calculate your specific costs with our free calculator.