3 Signals from NAB Show 2026 and What Intelligent Observability Delivers

Last week, 58,000+ media and entertainment professionals gathered in Las Vegas for NAB Show 2026. From new agentic workflows reshaping broadcast operations to AI-powered content tools on nearly every booth. One gap dominated my conversations: everyone was talking about building AI, but nobody was talking about operating it. 

When I took the stage in AWS’ Theater for my session, From Monitoring to Action: Powering Agentic AI and 2-3x ROI in Media & Entertainment, seats filled up fast. The topic wasn’t AI for content creation, generative AI for editing, or synthetic media. It was about how organizations operate AI, from observing, governing, and scaling. I met with leaders from major streaming platforms, broadcasters, and sports rights holders, and they all expressed an appetite to operate AI fast.

NAB Show had double the AI exhibitors compared to last year, including two dedicated AI pavilions. The streaming landscape has made a shift from growth-at-all-costs to one defined by operational efficiency and viewer retention. The hard question is how you make all of it function reliably and profitable at scale, with AI layered on top. Here are three signals we saw:

1. AI is Everywhere, But Nobody’s Talking About Operating AI

Walk the NAB Show floor and you could not escape AI displayed on banners. There was automated captioning, AI-assisted editing, synthetic media, AI-powered ad break identification, and more. Having been to many NAB Shows, I can say this is the biggest shift the industry has seen. 

Nearly all of the AI conversation was about AI for content, from creating it, tagging it, personalizing it, to monetizing it. Very few companies were talking about how you actually run AI workloads, how you know if your AI pipeline is healthy, how you catch hallucinations, manage token costs that spike during a live event, how you diagnose latency added by a RAG system, or how you observe agentic workflows you’re building before they create more problems than they solve. 

New Relic was one of the only companies at NAB with a message around AI for operational intelligence, and my session confirmed the interest. I structured it around two connected stories: AI for observability where intelligent agents surface operational insights across your media stack, and observability for AI where visibility into whether your AI workloads are performing. The transition is from manual war-room troubleshooting to autonomous predictive remediation that resolves issues before they impact viewers.

New Relic AI Monitoring delivers real-time visibility into the full AI stack from model performance, token costs, and response tracing that shows the complete chain from prompt to tool execution in a single view. This includes shadow AI models running in production without your knowledge. During our cabana meetings, I showed the output of New Relic's agentic intelligence running against a customer's production environment. You could see actual insights from their telemetry. 

2. AI is Making Industry Pains More Expensive to Ignore

The fundamentals of what makes media and entertainment operations hard haven't shifted. Quality of Experience (QoE) degradation like buffering and start failures still kill viewer retention and ad revenue. Live events still create the highest-stakes infrastructure challenges in the industry. And diagnosing whether a streaming issue originates at the Content Delivery Network (CDN), origin server, encoder/packager, or client device is still harder than it should be.  Teams still need to get to the ‘why’ something is happening and not ruminate on the “what”.

The cost of getting it wrong is now higher. According to our 2025 New Relic Observability Forecast, a high-impact outage now costs nearly half of all media companies over $2 million per hour. That figure reframes what golden metrics such as startup time, buffering ratio, playback errors and ad fill rate means for the business. When any one degrades during a live event, you're losing the economics that justify the entire operation.

What's changed is the new layer of complexity. As media organizations deploy AI workloads, such as recommendation engines, personalization systems, and agentic workflows, they're adding new failure surfaces that traditional monitoring tools weren't designed to catch. An AI system can be running and still be delivering inconsistent responses, burning through your token budget, or introducing latency in the stack before any dashboard shows it.

New Relic's Intelligent Observability Platform covers the complete pipeline, from camera to glass, connecting content ingest, transcoding, CDN delivery, client-side playback, and the AI layer in a single view:

• Streaming Video & Ads Intelligence: correlates QoE metrics with application and infrastructure performance to pinpoint root cause across the full stack, giving teams the ‘why’ behind the ‘what’
• Engagement Intelligence: tracks user behavior to surface leading indicators of churn before subscribers cancel
• Predictive Intelligence: AI-driven anomaly detection that catches issues before they reach viewers

Accedo achieved a 64x improvement in streaming performance with New Relic. Condé Nast reduced MTTR by 60% and improved response times by 70%. These pains are expensive and solvable with full-stack visibility.

3. Point Solution Fragmentation is Reaching a Breaking Point

In my session, I described it as the fragmentation minefield, the common trap of juggling eight or more disconnected tools. What's changing is the cost of maintaining it, and AI is what's making that cost impossible to ignore.

As media companies map out agentic architectures, they’re looking at eight or more disconnected tools, each with its own data model and AI endpoint. This makes synthesizing useful answers more expensive, more token-intensive, and more complex to orchestrate than a single platform that already holds the data. AI amplifies the cost of fragmentation and the minefield gets bigger.

A customer story from the week illustrates where the market is heading. A media company that had relied on a dedicated player analytics QoE point solution for over a decade, recently evaluated New Relic on their own. In four weeks, they validated the data model, confirmed their requirements, compared the economics, and made the switch. You know the market has shifted when a ten-year embedded relationship ends because a unified platform is simply more capable and more cost-effective.

New Relic gives media companies one platform for the entire stack including APM, infrastructure, DEM, streaming video, logs, and AI monitoring. One platform, one data model, one place to look when something goes wrong.  Now that we can unleash the power of media-skilled agents on this data, we are finally “flipping the script”.

What This Means for Your 2026 Strategy

Last year, AI was aspirational. This year, VPs were telling us they'd been personally running agent experiments for months. The industry has moved past asking what's possible and is focused on what works in production. Three things worth acting on:

Start observing your AI workloads now. Know what your models are doing, what they're costing per request, and where latency is accumulating, before your viewers or your CFO surfaces the problem.

Don't add AI complexity to an already fragmented stack. The core industry pains haven't gone away. Adding AI workloads on top of numerous disconnected and isolated expensive tools creates new failure surfaces without solving the existing ones. Full-stack visibility, from ingest to playback to the AI layer, in a single platform is what makes both manageable.

Do the consolidation math before you design your agentic architecture. Token costs, integration overhead, and engineering time add up faster than most teams expect. If a ten-year embedded point solution can lose a customer in four weeks because the economics no longer hold, the same calculation is available to every team still running multiple tools.