Operations | Monitoring | ITSM | DevOps | Cloud

GPU observability isn't one thing - it's eight connected layers from silicon to cost. See why correlation across layers is what cuts debugging from 2 hours to 2 minutes, and why most teams instrument only one or two.

Almost every production application uses a number of different tools and libraries,whether that’s a library to communicate with a database, a cache, or frameworks like Nest.js or Nitro. To be able to observe what’s going on in production, application developers reach out for Application Performance Monitoring (APM) tools like Sentry. But there’s an inherent problem: the performance data that APM tools need is most often not coming natively from the libraries themselves.

GrafanaCON 2026 kicked off in Barcelona, which is a fitting city to reveal the latest updates in Grafana 13. In 2013, Grafana Labs Co-founder Torkel Ödegaard made the first commit for what would become Grafana while he was on vacation in the Catalan city. "I was traveling here for the Christmas holiday and I got a cold and spent most of the day in bed coding and working on Grafana," said Torkel during the opening keynote of GrafanaCON, our biggest community event of the year.

The observability industry has developed great tools for using metrics, logs, traces, and profiles to monitor the cloud native applications that have dominated the last decade of software development. But when it comes to understanding what an AI system is actually doing, we’re often left reading raw conversations, guessing at quality, and reacting too late. And that’s a problem.

For years, infrastructure stability could be approximated through static limits. If CPU utilization exceeded a defined percentage or response time crossed a fixed boundary, risk was assumed to increase in a predictable way. Monitoring systems were designed around that assumption, and for contained environments, it largely held true.

AI Classify reaches general availability this month, TDM gets a Docker image, and we’ve added a few things you’ve been asking for.

Make AI do the dirty work while you focus your energy on strategy. CloudZero's Ryland Bowles shows you how. Every FinOps engineer is worried that AI is going to steal their job. I’ve worried about it. But I’ve also experimented extensively with AI, and I’ve got a pretty clear sense of what it can and can’t do in a FinOps context.

Claude Opus 4.7 holds at $5/$25 per million tokens — but a new tokenizer inflates costs up to 35% on identical text. Here's what Opus 4.7 actually costs at production scale, how it compares to Sonnet 4.6, and the six levers that determine where your bill lands.

Developers and SREs who rely on Microsoft Azure DevOps often face fragmented workflows when investigating issues or reviewing code quality. Troubleshooting an error can require jumping between observability tools and source code repositories as you manually connect traces, stack frames, and commits. At the same time, security vulnerabilities, misconfigurations, and flaky tests may go undetected until later stages of the software delivery life cycle (SDLC), where they are more costly to fix.

UK organizations are increasingly required to design systems that account for data residency requirements, ensuring that operational data remains within national boundaries. Many teams already run their applications on AWS infrastructure in the UK, but telemetry data can still be processed outside the region, creating gaps in visibility. Datadog’s upcoming UK availability zone solves this by keeping telemetry data in the same region as the workloads that generate it.