Inside the AV Cloud Summit 2025: AV Enters the Physical AI Era

The Fall 2025 AV Cloud Summit, sponsored by Xyte, opened with a clear message: AI has already moved from concept to operational reality in AV, and the industry is now sorting out how to harness it. Across manufacturers, enterprise IT, networking, and investment, the consensus was unmistakable – AI is already embedded in the systems we deploy, the workflows we support, and the expectations end users bring to the table.

A diverse group of leaders shaped the discussion. Intel Capital’s Avi Bharadwaj explained how investment decisions increasingly depend on data scale and repeatable use cases. Prateek Srivastava, who led enterprise AI at Dell, described what it takes to operationalize AI inside large organizations. Shure CTO Sam Sabet shared real examples of systems adapting automatically to spaces and users. NETGEAR’s Pramod Badjate highlighted the critical role of telemetry and edge intelligence. And EY’s Ron Sklaver, speaking for the enterprise buyer, grounded the conversation in real-world usage, support pressure, and measurable outcomes.

Guiding the entire session, Gary Kayye, founder of rAVe kept the panel focused on actionable insights, cutting through the hype to zero in on what AV and IT teams actually face in the field today.

To open the session, Xyte CEO and cofounder Omer Brookstein introduced the idea of physical AI – the convergence of AI models in the cloud with real-world, network-connected devices, making them more like platforms rather than static hardware. Platforms are always connected and up-to-date while continuously improving, and can unlock new features over time. He posed the question that framed the broader discussion: “What if every microphone, every speaker, every control system could be like my Tesla?”

The Core Challenge: Devices and Data That Still Operate in Isolation

Despite AI’s rapid advancement, the panel agreed that most AV systems remain fundamentally disconnected from the intelligence now flowing through modern software ecosystems. Physical AI requires a foundation that many AV products simply weren’t built for, Brookstein said, describing a structural mismatch between how devices are built and how modern systems operate.

“All the physical devices around us... they’re completely disconnected from the [AI] revolution,” he said. In most AV environments, support still depends on people. OEMs assign thousands of employees to resolve common tickets, many of which repeat across deployments. Brookstein pointed to an average cost of $22 per service call. Yet still, most teams try to scale by expanding headcount.

NETGEAR’s Pramod Badjate emphasized the urgency: “AV deployments are more demanding than regular networks. We hear this from integrators constantly.” Automation, he said, has to include the infrastructure itself - not just the endpoints.

That gap between what devices report and what teams can act on is where the next stage of the summit turned, focusing on how AI assistants reshape support from the inside out.

AI-Driven Support: Shifting the Model from Reactive to Autonomous

Support in AV environments is evolving from a ticket-based workflow into an intelligent, embedded layer within the system. As devices become continuously connected and generate richer telemetry, AI can observe behavior, detect drift, and intervene before issues escalate. Instead of waiting for help-desk tickets, the system itself tracks changes, identifies emerging faults – such as outdated firmware or misconfigurations – and resolves them in real time.

This reflects what today’s end users increasingly expect. EY’s Ron Sklaver captured this mindset directly: “I love the idea that AI can… notice [an issue] and just fix it… I don’t even have to know what happened.”

Shure CTO Sam Sabet outlined what this looks like inside a room. “You need the eyes and ears in the space,” he said, emphasizing how aligned intelligence steadies the environment as it runs. Support becomes a quiet part of the room’s operation, carried by the same awareness that keeps everything stable.

The Infrastructure Imperative

Intelligent AV depends on infrastructure that can participate in the same AI-driven loop that guides device behavior. NETGEAR’s Pramod Badjate outlined why this foundation matters, pointing to the need for networks that can adapt as conditions shift across rooms and workflows. Stable latency, clear telemetry, and consistent visibility give automated agents the context they need to intervene early and keep systems predictable.

EY’s Ron Sklaver described what that looks like from the user side. “It’s telling us something and it’s allowing us to have much different conversations with facilities, with real estate, even with the service lines,” he said, referring to the insights his team extracts from room-level data. His point underscores the role of infrastructure in every AI-driven outcome. The network supplies the signals that guide adjustments, inform support, and keep environments aligned without additional effort.

Where Capital Flows

The conversation naturally shifted from the foundations enabling AI-driven systems to the forces shaping who builds and funds them. Investors watch the same signals that engineering teams depend on: clean data paths, environments that adapt through real usage, and platforms designed to support automated decisions at scale. 

Intel Capital’s Avi Bharadwaj described how those conditions translate into long-term value. He explained his approach clearly: “One rubric that I use is to see what types of use cases have low complexity, high repetitiveness and low cost of failure.” That criterion shapes where capital moves, favoring platforms that solve problems felt across many environments rather than narrow, isolated needs.

As Dell’s former Enterprise AI leader Prateek Srivastava explained, legacy enterprises are still burdened by slow modernization cycles, but the next wave of companies won’t operate that way. “The new enterprises which people are going to create, they will be AI first,” he said, pointing out that these businesses can move faster, automate more, and make decisions in minutes rather than months. For investors, this creates a clear signal: capital will gravitate toward platforms and ecosystems that enable this AI-native operating model – whether through real-time telemetry, automated support, or intelligent device networks.

What the Market Wants Now

EY’s Ron Sklaver offered a clear window into current buyer expectations. He relies on real usage data to understand how rooms behave, and he wants systems that act on that information without adding work. “It’s telling us something and it’s allowing us to have much different conversations with facilities, with real estate, even with the service lines,” he said, describing how room-level signals now shape decisions across the organization. His expectation is simple: environments should stay predictable, reveal issues early, and avoid dragging teams into recurring fixes.

Pramod Badjate sees the same need from the network side. Integrators ask for deployments that hold their configuration, adapt to mixed protocols, and keep performance steady without constant adjustment. 

Building Together

The closing conversation focused on how progress moves when every contributor strengthens the same foundation. Device makers supply the data that reveals behavior in real spaces. Networks keep that information moving with enough precision for automated systems to act on it. Cloud platforms shape those signals into guidance that teams can use. Enterprises set the priorities that define what matters most in the environment. Omer Brookstein captured the momentum behind this emerging collaborative, AI-powered ecosystem, noting, “I really do think 2026, 2027 are going to be amazing years for the industry as we finally leverage the power of this cloud and AI tools.”

To see the full discussion and hear each perspective in detail, watch the complete AV Cloud Summit Fall 2025 session here.