New Relic has launched Infrastructure 360, a new interface on top of its existing infrastructure monitoring. It maps cloud resources, dependencies and configuration changes. The problem it addresses is real. The statistic it leads with is out of date, and the AI observability market it is selling into has shifted under it.
What New Relic launched
Infrastructure 360 gives platform teams an end-to-end view of their cloud estate. The tool agentlessly discovers AWS and Azure resources and flags where metrics coverage is missing. It tracks configuration changes down to property-level diffs. Dependency mapping runs across Kubernetes, networks and gateways.
It is built on New Relic’s existing infrastructure observability capability rather than replacing it. The new part is the interface and the depth of the dependency and change data.
It is available to New Relic customers in public preview from 13 October. That is a preview, not general availability.
The outage statistic is out of date
The release says “half of major production outages stem directly from these untracked infrastructure changes”. It cites no source, and the most recent data tells a more nuanced story.
The Uptime Institute’s 2026 Annual Outage Analysis puts configuration or change management at 41% of major network outages. That is down nine percentage points in a single year. For IT system and software outages, firmware and software faults now account for 53%, ahead of configuration and change management at 51%.
Earlier figures were higher. Uptime’s 2025 survey put change and configuration issues at 62% of major outages, and New Relic’s own 2024 Observability Forecast attributed 55% to change.
The trend is not that change failures have vanished. It is that change discipline works on the changes teams can see, and the risk is moving deeper into the stack. A static “half” understates how much that boundary has moved.
Why AI observability is now a requirement
The reason infrastructure visibility matters more now is that AI workloads scale differently. They consume compute and memory unpredictably, they change underlying resources at speed, and they fail in ways that classic host metrics do not capture.
Gartner’s July 2026 Magic Quadrant for Observability Platforms names AI observability as an emerging requirement, driven by LLM, GenAI and agentic AI adoption. It predicts that 40% of organisations deploying AI will use dedicated AI observability tools by 2028. It also expects LLM observability investment to rise to 50% of GenAI deployments, from about 15% today.
Gartner is also blunt that many vendor claims about autonomous operations remain ahead of reality. New Relic’s own Autopilot — its set of SRE agents — sits in exactly that contested space. Infrastructure 360 is positioned as the data foundation that makes those agents reason better.
The real pressure is cost
Here is the part the release underplays. The observability market’s binding constraint is no longer detection. It is spend.
A survey of 500 US technology professionals by groundcover found that 49% say AI workloads account for 26% to 50% of their observability costs. Some 39% spend $1 million to $5 million a year on observability, and 53% exceeded budget by 10% or more.
The drivers are specific: high-cardinality metrics and excessive ingestion, higher-fidelity telemetry for AI systems, and overlapping tooling. Those are the same fragmentation problems Infrastructure 360 is designed to address, which makes the product’s cost argument as strong as its visibility argument.
Gartner puts the market at $14.3 billion by 2028. It notes that 5% of its clients already spend more than $10 million a year with a single provider. It also warns that OpenTelemetry and eBPF have commoditised data collection, so vendors must differentiate on analytics and automation rather than on ingestion.
What that means for the launch
Infrastructure 360 competes on exactly the right axis. If you cannot tell which change broke production, you cannot cut the cost of finding out. Correlating a config diff with a latency spike is a genuine time saver, and time is the metric that maps to money.
But the release sells visibility while the market is buying savings. Uptime’s 2026 data adds the other half. 57% of operators said their most recent major outage cost more than $100,000. One in five crossed $1 million.
A product that shortens incident response is a product that reduces outage cost. New Relic makes that case implicitly rather than explicitly, and gives no number to support the “drastically reduce” claim.
What to watch
Whether Infrastructure 360 reaches general availability on a stated date, and whether New Relic publishes response-time or cost-reduction figures from preview customers. Whether the AI observability category consolidates around full-stack platforms or splits into specialist model-monitoring tools. And whether telemetry cost, not capability, becomes the deciding factor in renewals.
The visibility problem is real. The market, right now, is pricing the bill.
Editor’s Note
This analysis draws on New Relic’s Infrastructure 360 release and product documentation, New Relic Autopilot documentation, the Uptime Institute’s 2025 and 2026 Annual Outage Analyses, Gartner‘s 2026 Magic Quadrant for Observability Platforms and its AI observability forecasts, a groundcover survey of 500 US technology professionals conducted by Atomik Research, and New Relic’s 2024 Observability Forecast. Where a figure is a vendor’s own or an estimate, we say so.