New Relic's 2026 Observability Forecast, produced with Enterprise Technology Research (ETR), surveyed 2,575 IT and engineering leaders and practitioners worldwide, up from 1,700 last year, and the finding at the center of nearly every result is the same one: the tools that made software faster to ship are also the reason it's getting harder to see.
For the second year running, AI is the single biggest force pulling organizations toward observability. It's not one driver among several anymore. Adoption of AI applications, security and governance pressure, and the rise of agentic AI systems now top the list of what's driving demand, and all three trace back to the same root cause: AI is moving faster than the infrastructure built to watch it.
The gap nobody's closing fast enough
Autonomous AI agents are already in production at most organizations surveyed. Only 25 percent of organizations report they haven't deployed or don't plan to monitor agentic AI at all. The problem is what's happening inside that remaining 75 percent. A full quarter of organizations have already put agents into production with no monitoring in place whatsoever. That's a live production system, making decisions and taking actions, that nobody is watching right now.
Code review can't keep up with code generation
The same tension shows up in how software gets written. Two out of three organizations say AI now generates or substantially rewrites more than half of their code each week. Eighty-three percent of leaders agree that this makes reliable observability more critical than ever. When a machine writes the majority of your codebase, the only reliable way to know what it's actually doing is to watch it run.
Outages cost less. They still happen constantly.
The average organization now loses about $74 million a year to high-impact outages, down slightly from $76 million last year. While not a significant change year-over-year, it tracks with faster detection and resolution times industry-wide. But don't mistake a smaller bill for a smaller problem. More than a third of organizations still experience a high-impact outage weekly or more, and the share hitting one multiple times a day has roughly tripled year over year. The cost curve is bending. The frequency curve isn't.
The tools are multiplying again
After two straight years of consolidation, that trend is reversing. The average organization ran ~4 observability tools in 2025, but is trending back up to 5 this year, likely because AI adoption is pulling in a wave of new, narrow, AI-specific point tools. Half of organizations still say they'd rather run a single consolidated platform than juggle several. The ones who actually manage to do that, instead of adding one more dashboard every time a new AI capability shows up, are the ones this report shows pulling ahead on every other metric.
What this means going into 2026
Taken together, the pattern is consistent: organizations that are watching what they've built (their agents, their AI-generated code, their production systems) are moving faster and paying less for it. Organizations that are shipping without watching are accumulating risk that hasn't shown up in the numbers yet, but presents a significant runtime risk moving forward.
The full 2026 Observability Forecast breaks down all of this in far more depth: how AI is reshaping the incident lifecycle, why OpenTelemetry adoption is accelerating, what the highest-performing organizations are doing differently with their tool strategy, and where the biggest blind spots remain.
Download the full report to see where your organization stands.
Jim Young is the Director of Content Marketing at New Relic, where he leads the company's content strategy and thought leadership initiatives. With a background in content marketing and brand strategy from his 5+ years at Twilio and Segment, he is particularly focused on exploring the impact of AI on the observability space.
The views expressed on this blog are those of the author and do not necessarily reflect the views of New Relic. Any solutions offered by the author are environment-specific and not part of the commercial solutions or support offered by New Relic. Please join us exclusively at the Explorers Hub (support.newrelic.com) for questions and support related to this blog post. This blog may contain links to content on third-party sites. By providing such links, New Relic does not adopt, guarantee, approve or endorse the information, views or products available on such sites.
