Verizon Business positions its network assets to support AI

Enter AI Connect

With the unveiling of its Verizon AI Connect platform, an integrated suite of solutions and products designed to enable businesses to deploy artificial intelligence (AI) workloads, the service provider has positioned itself in the market segment for a host of customer profiles--hyperscalers, cloud providers and global enterprises.

One key trend Verizon hopes to help its customers with is training current advanced AI models. This concept, known as inferencing, is poised to drive new demands for additional computing power.  

According to McKinsey, 60-70% of AI workloads are expected to shift to real-time inference by 2030, creating an urgent need for low-latency connectivity, computing, and security at the edge beyond current demand. 

Verizon Business’ strategy to fulfill new AI demand is to approach it with what it says are “reimagined existing assets integrated into Verizon’s intelligent and programmable network.”

Designed to deliver AI workloads, Verizon will leverage its array of wireless and wireline network capabilities, including the macro 5G network and high-speed fiber connectivity, to edge compute environments and space, power, and cooling infrastructure. Google Cloud and Meta are leveraging additional capacity to support their AI workloads.

“Verizon AI Connect is the name of our strategy and suite of offerings intended to meet the growing demand for AI applications from our ecosystem partners and end-user customers,” Malady said. “It's a vision that allows us to utilize existing assets in new ways to service this technology revolution. Whether the AI workloads are across a multi-cloud environment deployed on our customers' premises or at the edge of the networks, we have a suite of offerings that businesses can utilize to leverage AI capabilities fully.”

The foundation of Verizon’s AI drive is centered on connectivity and edge computing.

From a wireless perspective, Verizon’s partnership with NVIDIA is focused on reimagining how GPU-based edge platforms can be integrated into its 5G private networks. 

“Our vision is built around best-in-class connectivity and edge compute assets,” Malady said. “It all starts with connectivity.”

Verizon is working with NVIDIA to reimagine how telco functions can work along with AI workloads.

‘We are starting this development at the far Edge in a private network,” Malady said. “Its applicability will eventually extend beyond that and potentially into the macro network.”

Deep fiber reach

While wireless is a cornerstone of its connectivity options, Verizon Business can also draw on its well-entrenched business fiber assets.

Verizon Business' fiber assets developed over several decades include the 71-city One Fiber buildout and several long-haul networks. These provide light and dark fiber services, edge-to-cloud connectivity, and network programmability.

Today, Verizon's fiber network spans over 500,000 route miles and connects over 51% of its wireless cell sites. 

“Over the last twenty years, we have built unmatched fiber assets, not only in our ILEC footprint via Fios but also outside through our 71-City One Fiber build-out,” Malady said. “With our long-haul network that stitches these metros, we can provide light and dark fiber services to our customers over a large geographic area.”

However, the long haul and metro networks are only part of the enterprise AI equation. The service provider also has an extensive on-net and near-net fiber business footprint to serve enterprises. 

Joining AT&T, Verizon is ranked second on Vertical Systems Group’s year-end 2003 U.S. Fiber Lit Buildings LEADERBOARD. Since the base of commercial buildings with two or more fiber providers, including the growing presence of large cable MSOs like Comcast and Charter, continues to increase, Verizon’s standing could be challenged.

Nevertheless, Verizon stands in a unique class of providers. To qualify for this benchmark, Verizon must have 15,000 or more on-net U.S. fiber-lit sites, including commercial buildings and data centers, by year-end 2023.

“We believe Verizon will serve as the essential link between these Gen AI applications and the firms who utilize them,” Malady said. “With a growing network of over 16,000 near-net enterprise locations, Verizon is positioned at the forefront of AI deployment, providing the connectivity solutions that will enable the seamless integration of AI applications across various industries and sectors.”

Supporting edge data centers

Data centers are also another key element in Verizon’s AI strategy.

While Verizon sold off a significant portion (24 sites) of its data center sites to Equinix in 2016, the company still owns 20 data centers, of which 17 are its own, and another data center operator operates three.

According to Verizon, it also connects to a large community of third-party data centers, ranking nearly a third in the market.

“Data centers will continue to be crucial in this new AI-powered world, and 40% of data centers are expected to face operational constraints by 2027, another factor in workloads moving to the Edge,” Malady said. “Verizon's extensive fiber and programmable networks are ready to meet these demands across key markets.”

Verizon is certainly not alone in pursuing data center opportunities with fiber assets. Fellow ILEC Lumen signed an agreement with Corning that reserves 10% of the fiber manufacturing company’s global fiber capacity for the next two years to interconnect AI-enabled data centers. It will serve Lumen’s U.S. intercity network, which includes diverse routes to more than 50 major cities nationwide.

Malady acknowledged Verizon is “not alone in recognizing the opportunity.”

“Industry analysts are seeing and predicting further significant growth in the AI market,” he said. “Hyperscalers are expanding into new markets, and they need high-capacity connectivity. The industry is seeing a surge in cloud migration and colocation demand.”

However, fiber is only part of Verizon's support mechanism for hyperscalers.

The provider can also offer power, space, and cooling. Verizon said it has 2 to 10 megawatts of usable power across its sites.

“Power, space, and cooling are the currencies in demand right now, and we have all three,” Malady said. “We have facilities across the United States that either have spare power, space and cooling or can be retrofitted.”

Also, Verizon has a wide array of land holdings. The company currently has between 100 and 200 acres of undeveloped land.

Malady said, “Some of this land is currently zoned for data center build and much of it in prime data center-friendly areas.”

AI customers rising

All of Verizon’s assets would not have much meaning without real customers.

Today, the service provider has over $1 billion in funnels, leveraging its existing network and related assets.

“Major players such as Google and Meta have purchased capacity in our network to use it for their AI workloads,” Malady said. “Some of these deals are reflected in our fourth-quarter results and contributed to the margin improvements you saw in the quarter.”

Verizon is also enhancing its partnership ecosystem to target new AI customer opportunities.

One of its partners is Vultr, a global GPU service provider and cloud computing provider. Vultr will deploy its GPU as a service infrastructure in one of Verizon’s data centers and leverage its fiber network for distribution. 

Verizon will levearge Vultr's capabilities to expand its AI support capabilities and reach further.

Malady said that this partnership will be applied to accommodate other needs. “Over time, we anticipate helping to broaden their reach and enable our mutual customers with AI training and inference capabilities at the Edge,” he said.

Complementing fiber network capacity, Verizon Converged Intelligent Edge Network will allow customers to control where and how AI workloads run. Verizon's "Converged Intelligent Edge Network" is the telco’s advanced network infrastructure that integrates edge computing capabilities. This allows for real-time data processing closer to where it's generated, minimizing latency and enabling applications like AI and IoT with high bandwidth and low delay, leveraging the provider’s wireline and wireless networks to deliver a unified, intelligent edge experience across different platforms. 

“Our investments in our Converged Intelligent Edge Network over the last several years will pay dividends in this new world as users need even more bandwidth and network visibility,” Malady said. “Customers want to control where and when AI workloads are running, so we are building new capabilities to give users more insight and control via new tools.”

For related articles, visit the Optical Tech Topic Center.
For more information on optical components and suppliers, visit the Lightwave Buyer’s Guide.
To stay abreast of optical communications technology, subscribe to Lightwave’s Enabling Technologies Newsletter.