more distributed processing across regional facilities and even consumer devices. Google, he said, is already developing both ends of that spectrum, from enormous AI data centres to smaller open-weight models capable of running directly on laptops and smartphones. “The models are getting better and they’re getting smaller. We’re seeing AI moving from massive centralised infrastructure back towards distributed computing on people’s own devices.” The panel concluded that edge computing is no longer simply about where servers are located. Instead, it is becoming an intelligent fabric that dynamically places workloads wherever they deliver the greatest benefit. As Nock observed throughout the discussion, the industry’s challenge is no longer building ever more sophisticated technology, but ensuring that all of it ultimately improves one thing: what the viewer sees on the screen.
As services expand across dozens of platforms, operating systems and software versions, failures often occur after streams have technically started, making them invisible to conventional monitoring systems. To address this, Witbe is using AI to test services on real consumer devices using natural language instructions instead of traditional scripted automation. “We don’t write scripts anymore. We express an intent in plain language and let the AI understand what it’s seeing on the screen. The system adapts instead of breaking every time an interface changes.”
After showing a clip from The Wizard of Oz, he compared engineers to the wizard’s role as “the man behind the curtain”, building ever more sophisticated systems while forgetting where success is ultimately measured. “All this monumental infrastructure exists to serve one main endpoint – the screen of a real device. If it collapses on the last pixel, the entire workflow is perceived as a failure.” He said traditional monitoring often reports everything as healthy while viewers experience frozen pictures, black screens or unresponsive applications. “The problem isn’t a lack of data. It’s the gap between what we measure and what the viewer actually experiences on a real device.” Using the Apollo 13 mission as an analogy, El Gharbaoui noted that astronauts first recognised a critical oxygen leak simply by looking out of the spacecraft window rather than relying solely on telemetry. “In exactly the same way, the data can be blind while the human eye is not.” He argued that the streaming industry’s biggest challenge has moved away from network reliability towards the increasing complexity of consumer devices. “The pipes are solid. The number one challenge today isn’t CDN performance anymore – it’s playback across fragmented players and smart TVs.”
Chris Hampartsoumian,
Customer Solutions Engineer at Google Cloud, described edge computing as fundamentally about
distributing capacity.
“What the edge gives you is distributed capacity. Sometimes you need low latency, but increasingly it’s about having massive computing capacity exactly where it’s needed.” He suggested computing is entering another cycle, moving away from highly centralised cloud infrastructure towards
Made with FlippingBook - Online magazine maker