Building the Measurement Layer Livepeer Needs: Introducing NaaP Analytics

By the Cloud SPE Team — Livepeer.Cloud

The Network Is Growing. Now It Needs to Be Measurable.

Livepeer has evolved from a decentralized video transcoding protocol into a full-fledged AI compute network — handling image generation, video synthesis, LLM inference, text-to-speech, and more. Orchestrators around the world power these workloads on GPUs ranging from GTX 1080s to RTX 5090s serving gateways and applications that depend on performance and reliability.

But here's the gap: until now, there hasn't been a shared, network-wide way to measure how well it's all working.

Gateway providers couldn't easily compare orchestrator performance across regions or workloads. Orchestrators had no standardized way to demonstrate reliability. Developers evaluating Livepeer for production use had to trust marketing materials instead of transparent data. And the broader ecosystem lacked the foundation needed to build SLA-aware routing, scaling, or any kind of production-grade service guarantees.

That's what this project changes.

What We Built

The Cloud SPE has joined forces with the Livepeer Foundation, Livepeer Inc, and the Livepeer Community to design, build, and deploy the NaaP (Network-as-a-Product) Analytics MVP — a complete metrics, analytics, and observability platform for the Livepeer AI network.

Funded through the Livepeer Treasury and aligned with the Livepeer Foundation's roadmap to make network data more observable, this project delivers:

Core SLA Metrics

A standardized set of performance, reliability, and demand metrics — covering everything from job success rates and GPU throughput to latency, FPS stability, and payment economics. These metrics are sourced from gateway telemetry, orchestrator signals, and reference load tests, unified into a single analytics layer.

Network Test & Verification Signals

We operate reference load-test gateways that generate consistent, reproducible performance signals across live AI pipelines. Public test scenarios are designed to reflect real workloads, are transparent and community-verifiable, and feed directly into the same analytics layer as organic network traffic.

Analytics & Aggregation Layer

A purpose-built data pipeline transforms raw events into network-level views. Events flow from Kafka through ClickHouse, normalized through a resolver service, shaped by dbt semantic models, and served through a Go REST API. The architecture prioritizes efficient querying — dashboards never need to scan raw job data.

Public Dashboard & APIs

A standalone Grafana-powered dashboard presents live and historical metrics across four key domains: system health, real-time operations, economics and payments, and performance drill-downs. Public, read-only APIs expose aggregate SLA scores, GPU supply inventory, and orchestrator leaderboard data. Gateways and ecosystem teams can consume this data directly or mirror it into their own systems.

Operations & Stewardship

The entire platform is deployed across a production-grade infrastructure with automated monitoring, Prometheus metrics, and documented runbooks. We're committed to maintaining and operating this infrastructure for the Livepeer community.

Why This Matters

This isn't a dashboard project. It's the measurement layer that the Livepeer network requires to evolve from "decentralized compute that works" into "production infrastructure you can bet your business on."

For Gateway Providers: You can now evaluate orchestrator performance with real data — compare reliability, latency, and throughput across regions and workloads before routing a single job.

For Orchestrators: Your performance is now visible and comparable. Good work gets recognized. The network rewards reliability, not just availability.

For Developers & Partners: Evaluating Livepeer for production use no longer requires a leap of faith. The data is public, the APIs are open, and the methodology is transparent.

For the Ecosystem: Future SLA-aware routing, automated scaling, and production service guarantees all depend on trusted measurement. This is that foundation.

The Road Here

This project didn't arrive overnight. It started with community conversations about what Livepeer needed to become a true production platform. An initial pre-proposal in October 2025 drew thoughtful feedback — the community pushed back on scope, cost, and architectural complexity. They were right.

We listened. We reset. We narrowed the scope, reduced the budget, simplified the architecture, and prioritized time-to-value. The revised proposal earned broad support — from the community, from Livepeer Inc, and from the Livepeer Foundation — and passed the treasury vote in January 2026.

Work had already begun in November 2025, and we've executed across three milestones: metrics collection and aggregation, test signals and derived analytics, and stabilization. The result is a production system that's live, documented, open source, and ready for the community to build on.

What Comes Next

This MVP establishes shared measurement. It does not enforce SLAs, modify protocol incentives, or introduce routing logic. Those are future decisions that the community can now make with data in hand.

What this enables:

• SLA-aware routing — gateways can route jobs based on demonstrated performance, not just price
• Network quality scoring — aggregate reliability metrics that make Livepeer legible to enterprise buyers
• Data-driven governance — treasury proposals and network decisions grounded in observable outcomes

We built this as neutral, public infrastructure. The data belongs to the network. The code is open source. The APIs are public. And we're here to maintain it, improve it, and support the teams that build on top of it.

About Cloud SPE

Cloud SPE is a Special Purpose Entity within the Livepeer ecosystem, founded by three Livepeer orchestrator node operators: Speedy Bird Technologies, Mike Zupper (Xode App), and Papabear (Solar Farm). We operate free-to-use Livepeer gateways, build open-source tooling for the network, and work to make decentralized video and AI infrastructure accessible to everyone.

This is our third treasury-funded project. Previous work includes the original SPE gateway and demand generation infrastructure and the AI Performance Leaderboard integrated into the Livepeer Explorer. Each project has built on the last, and NaaP Analytics represents the most ambitious — and most important — step yet.

Join the conversation in Livepeer Discord.