At a PTC panel in Hawaii last month, Verizon and industry peers discussed how AI is reshaping networks and data centres, prompting the US carrier to outline its strategy to leverage dense fibre and private 5G for enterprise AI workloads.
In sum – what to know:
Edge infrastructure – dense metro fibre and private 5G will support enterprise AI inference close to where data is generated.
Strategic investment – current AI builds are based on real demand, the firm says, and its One Fiber project is boosting US cities.
Enterprise applications – the convergence of AI, IoT, and private networks is creating practical applications at the enterprise edge.
In the interests of writing shorter, here is a quick Verizon excerpt from a panel at PTC in Hawaii at the end of last month (January 19) – which purported to discuss ‘who actually wins’ out of networks and data centres (‘where AI meets’), but never actually provided an answer (except ‘everybody’, of course). Nevertheless, the assembled execs – also from data centre operators Equinix and DataBank, plus cloud provider Oracle and consultancy AlixPartners – managed a good discussion about how AI demand is reshaping networks and data centres, and how companies are managing risks, returns, and long-term strategies amid unprecedented capital intensity.
There were some good cliches, quotes, and analogies in there, always welcome: eggs in baskets (managing risk); sprints and marathons (infrastructure investments); science and art (capital deployments); repeats and rhymes (historical precedents); potholes and flat tires (operational missteps); white elephants (stranded assets); and cures for cancer (the upside of it all). We will perhaps write up some of these angles too, but for RCR (this writer), it was interesting to hear Verizon join the dots at the edge of the AI bubble – at the metro edge and the enterprise edge, around the supply of fibre and 5G as managed private slices and systems.
It made practical this new obsession with AI inference.
“This feels different,” said Jeffrey Hulse, president of network and partner solutions within the wholesale arm at Verizon Business. He was responding to a prompt about whether today’s AI build-out echoes the dotcom bubble 25 years ago and the cloud boom 10 years ago – and whether yet another round of speculative infrastructure-building risks leaving excess capacity when reality diverges from the original hype narrative. “It is not built on spec: putting tons of fibre in the ground, hoping people will buy bandwidth in the future. We were just oversaturated in the late ‘90s in that way. But [we now] have folks who pay their bills and want very-specific capacity.”
For well-adjusted telecom providers, Verizon sees growth not just from traditional core network demand, geared towards hyperscalers, but in dense and distributed connectivity in metro areas to support enterprise usage of AI. “It is compelling and it will fuel growth on the network side for the time to come,” said Hulse. “The thing that gets us excited is how you push AI out to the prem. Because you’re going to need deep dense fibre networks at the edge for enterprise, public sector, and for any one of the (GPU/compute) sled opportunities out there.” He referenced the firm’s so-called One Fiber project in the US to build a converged network for its wireless and wireline services.
The initiative dates back a decade, launching in Boston as a six-year undertaking in 2016 to replace legacy systems in the city and extend fibre for mobile backhaul, broadband, and enterprise services. It has since morphed into a multibillion‑dollar annual cap-ex commitment to densify the “top” 69 US cities with fibre”, and effectively incorporated acquisition of Frontier Communications in 2024. Hulse said: “We are really lucky. We started executing on a strategy to own as much of the macro cell-site environment as possible – so we could control the customer experience.” If you want to make a dotcom comparison, the fibre count is through the roof – he said.
“We used to put 12 fibres down a street, wall to wall. When we started One Fiber, we were putting 864 down Main Street [and are now] putting 1,600 on the most important routes – and the use case is to put all of our macro cells on there. Fifty-one percent of our sites are now on-net, managed with our own power and people, via our own NOC. Now we’re saying: how do we take those fibres, splice them, and go into big enterprises, small-and-medium businesses, and the public sector to give them private-line services, high speed ethernet, wave services. So we are well positioned: all that dense fibre is what we’re going to need for inferencing. We think we are ahead of this curve.”
So the focus is not just about wide-area coverage back to large data centres via mobile core networks and backhaul fibre systems, but also about the close-area edge, where data is generated and consumed – and where AI insights are inferred from AI models. Meanwhile, Verizon Business is making a decent fist of the supply of private 5G systems into enterprises, as covered in these pages historically, to connected IoT sensors of various sorts, to be paired with edge computing to run inference workloads at close quarters for short turnarounds. Amid all the grandstanding at PTC about big data centre projects and backhaul builds, it was good to hear Verizon Business bring it home.
Hulse commented: “Being able to connect with private wireless networks, and then pull all that bandwidth all the way back through inferencing – we’re incredibly bullish on what the future’s going to hold, and we think we’re incredibly well positioned for that space. We do not think it’s a bubble. Maybe some of the financing is a little circular in nature, where the same companies are all investing in the space. But from a network perspective, we’re very happy with where we are.”
It might be noted, for RCR readers that have followed the private 5G story for the last 10 years, that AlixPartners talked in the same panel about the AI effect to hyper-scale IoT (if it was ever in doubt), and Nvidia spoke in another session about its urgent focus on robotics in factories, and the like – where IoT sensing and 5G synthesis give direct reign to AI sensing. As we have written for years. So, it might be argued that, in ways, the whole tech industry is coming around to the IoT worldview, finally.
