Connect (X) 2026 Keynotes Deliver Past, Present, Future of Connectivity Industry

May 5: Verizon, Intel, Ericsson, JLL, Qualcomm, American Tower, NVIDIA, Nokia, T-Mobile Leverage Expertise to Predict Trends 

FORT LAUDERDALE, Fla., May 5, 2026 – Today at Connect (X) 2026 — the only business technology event in North America that unites the entire digital infrastructure industry — esteemed industry leaders united to reflect on, analyze, and project the course of the industry. WIA President and CEO Patrick Halley announced WIA’s launch of the Edge AI Infrastructure Initiative, in a cross-industry effort for edge AI infrastructure promotion and acceleration to anticipate the continued growth of AI at the edge.

Access photos here.

Notable quotes include the following: 

Welcome Keynote (transcript)

“In fact, WIA estimates that AI traffic now accounts for at least 4 percent of total wireless network traffic, representing nearly $3 billion in annual wireless network investment. We anticipate by the end of this year, around 10% of mobile data traffic will be driven by AI in some form.

“To make sure we are staying ahead of the curve, WIA has launched an Edge AI Infrastructure initiative, bringing together organizations from across our industry to promote the importance of edge AI infrastructure and actions to accelerate its implementation.”

— Patrick Halley, President and CEO, WIA

Verizon Keynote 

“We have today 40,000 sites that are running on what we call virtualized distributed units. Nobody else has that today. And we understand software, we understand the flexibility of it — that also helps all of us speed deployment…Verizon was also one of the first companies to deploy this infrastructure called mobile edge computing. We have it across 22 sites in the US today. It’s basically cloud computing right next to our packet core, so that when someone tries to look at a particular workload, maybe an app or something else, their activity does not have to go back to a central data center far away. It gets processed right next to our network. It has a significant [lower] latency and other benefits.” 

— Srini Kalapala, SVP and Chief Network Officer, Verizon

 

AI at the Edge: Where Compute Meets Connectivity 

This keynote panel explored how operators are turning mobile networks into AI‑powered platforms that drive efficiency today and lay the foundation for 6G.

“There are four primary workloads for AI inference: computer vision, small language models, video language models – which are taking the context out of the world videos – and the video language action model. Those are the [workloads] that are driving physical AI. All these workloads have different compute requirements. And when you are adding one more dimension… the number of cameras, the number of users and all those things… that’s where the compute requirements are different. What we are seeing right now in the market is pretty good visibility into what [enterprises] want to deploy now or what problem they want to solve. But they do not have good visibility or they are not thinking ahead of time. Like what they can achieve one or two years from now… future requirements are what drives the edge.”

— Bhupesh Agrawal, General Manager of Enterprise Edge AI and Private Wireless, Intel

“I think the largest driver of mobile edge computing (MEC) is actually that the scale of the data is vastly different today. If you look at the Ericsson Mobility Report, between 2023 and 2029 we see the data traffic is going to triple and that’s vastly coming from AI across the board. We also expect the uplink to grow by 10x by 2035 and we are seeing indications happening already today. All of this will be routing to some kind of a cloud. So what’s different today is the demand and it [being] generated by AI. Second, the networks today are built for time critical applications, which if you look at when we discussed MEC 10 years ago or so, it was best effort that we talked about at the time.”

— Joe Constantine, Chief Strategy and Technology Officer, Ericsson

“We need to carpet the planet with AI inferencing compute nodes. And it doesn’t always work in the general monolithic, multi-gig block data center campus. So, we’re in the infancy of this – this is still a research project – but we’re seeing all of our clients are scrambling to prepare their AI inference digital infrastructure. And it has to be done in these smaller nodes that are really close to the use case, which is AI inferencing and GPU. So, [in the] next five years, that’s all we’re going to see. And the edge is finally here.”
— Sean Farney, VP, Data Center Strategy – Americas, JLL

“Whether it’s the use of AI chatbots and AI,driven applications, that’s really changing the need for compute and the data that’s generated. Already 75% of data is generated at the edge and from a device perspective we’re moving away from downlink centric – watching videos or scrolling feeds to a more upload-centric AI generated data that increasingly needs the edge for compute-to-power latency improvements. I think in a couple years agentic data will overcome even the human generated data.”
— Dr. Ozge Koymen, VP, Technology, Qualcomm

“Edge is where physics collides with economics… if data gravity dictates that the cost of moving large data sets is either a performance problem or a cost problem, then the only answer is you have to move the thing closer to where it gets consumed. That is one version of Edge. That’s what’s propelled the existing data center industry to the point that it’s grown and will continue to grow the data center industry.

“If you think the current environment is giving you indigestion because things are changing too quickly, then we all need neck braces for the next couple of years is my prediction.”
— Jim Poole, VP, Product Development Edge Data Centers, American Tower

The AI-Native Network: Redefining the Future of Mobile Infrastructure 

This group of keynote panelists examined how Edge AI infrastructure is enabling low‑latency applications and rapid AI inferencing across wireless, fiber, and enterprise environments.

“With AI, we have the opportunity of rewriting wireless interfaces. Because now with AI, you can learn the RAN interface. You can put in so many great layers on layers, two capabilities that are AI-written. That’s where AI for RAN comes in. Whether it’s channel estimation, whether it’s beam forming, whether it’s scheduling, all of them are written with AI. So RAN as a workload itself is a perfect example of AI workload.

And then if you look into the future, you’ve got ISAC — kind of new 6G sensing applications coming in, perfectly suited for AI because these are multimodal inputs of RF, LiDAR cameras. All those inputs coming in to see, sense your environment and build other applications on top. So the moment you understand these workloads, you realize they’re inherently suited to be run on AI computing platforms. And hence the underlying infrastructure requires GPU. I must also add that it’s not only GPU, it’s always a trifecta of CPU, GPU, DPU that collectively come together to accelerate the workload and deliver the gains, the experience. Whether it’s latency, whether it’s tokens at scale, all those requirements. You need that accelerated computing infrastructure.”
— Kanika Atri, Senior Director – Telecoms, NVIDIA

“The more complex that model is with the billions of parameters – we talk about the machine learning models – then the complexity increases, then compute requirement increases. So I think this is where the combination would start to make sense. CPU plus GPU combination with the right mix of CPU and right GPU. Because we are not talking about GPUs on the distributed cell side which are, let’s say super power hungry or super expensive. We are talking about the small, tiny GPUs which are put in a PCIe card in a system. So it’s a different ball game you talk about. So that’s where the compute flexibility helps. Together with the CPU, it helps a lot, bringing more power to the compute, more efficient use cases. Even for RAN, I can say we are talking about more complex algorithms, which we would have never thought about before with available compute.”
— Aji Ed, VP, Head of Cloud RAN, SMO, and Network Management, Nokia

“Policy has to catch up. We need zoning and permitting that really keeps up with the pace of rapid deployment and efficient deployment… You talked about the advent of AI and the data growth explosion we’re going to see. We need more airwaves, more spectrum on the uplink, on the downlink. That is essential. With these things secured, we can unlock the full vision of physical AI.”
— Salim Kouidri, Senior Vice President – Engineering Operations, T-Mobile

 

May 6: Connect (X) Keynotes feature FCC’s Arpan Sura; AT&T; Digital Bridge and Tower Executives

Welcome Remarks

 

FCC Keynote

 

AT&T Keynote

 

Digital Bridge Keynote

 

View from the Tower

 

###

About Connect (X) by WIA

Connect (X) powered by the Wireless Infrastructure Association (WIA) is the only business technology event in North America that unites the industry stakeholders who deliver infrastructure solutions for ubiquitous connectivity. Connect (X) delivers market intelligence, networking and deal-making for the entire communications infrastructure ecosystem.

Media Contact: media@wia.org