Featured Story
A Nokia sale of mobile, especially to the US, would be nuts
Nokia's hiring of Intel's Justin Hotard to be its new CEO has set tongues wagging again about a mobile exit, but it would look counterintuitive and inadvisable.
Verizon's AI Connect scheme will stitch together data center colocation and edge computing resources, plus Verizon's fiber and 5G network infrastructure, to support AI workloads and the shift to AI inferencing.
Verizon is rolling out a suite of services to provide the power, space and cooling requirements, plus the network infrastructure needs, to support AI workloads.
Dubbed "Verizon AI Connect," the initiative brings together data center colocation and edge computing resources plus Verizon's network infrastructure. Among the network assets used to support AI Connect, Verizon plans to tap its metro and longhaul fiber deployments, ILEC footprint via Fios, 71-city One Fiber metro network buildout, lit and dark fiber services and macro 5G network.
"Verizon AI Connect is the name of our strategy and suite of offerings that are intended to meet the growing demand for AI applications from both our ecosystem partners and end user customers," explained Kyle Malady, CEO of the Business Group for Verizon, during the company's Q4 2024 earnings call.
Malady said AI Connect will use existing network assets to support AI workloads in multicloud environments accessed at the customers' premise or edge of the network. He also noted that Verizon has a sales funnel of "over $1 billion simply leveraging our existing infrastructure."
Malady said that Verizon's AI Connect initiative comes at a time of increased investment and demand for AI. Market analysts forecast over $1 trillion in investment in AI infrastructure over the next decade and AI network traffic could grow at a rate of 35%-plus Compound Annual Growth Rate (CAGR) over the next five years, he said.
"All of this AI infrastructure will need to be underpinned by secure network connectivity that will bridge the new distributed compute landscape."
Verizon's AI Connect partners and customers include hyperscalers such as Google Cloud and Meta, which are tapping Verizon's network infrastructure for their AI workloads.
Chipmaker giant Nvidia and GPU-as-a-service company Vultr are looking to Verizon to integrate their GPUs into Verizon's data center colocation infrastructure, explained Scott Lawrence, Chief Product Officer for Verizon Business. The service provider will support Vultr in achieving the "security, reach and scale that will help them as they grow their business," said Lawrence.
With Nvidia, Verizon will integrate their GPU chipsets into Verizon's private 5G service to deliver secure, on-prem services for enterprise customers deploying AI technologies, he added. Nvidia also partnered with Verizon in December 2024 to run AI applications and integrate GPUs over Verizon’s 5G private network with private Mobile Edge Compute (MEC).
Verizon plans to continue expanding its network resources for AI Connect and has between 100 and 200 acres of undeveloped land, some of which is already zoned for data center buildouts, said Malady.
Verizon Business' Lawrence said AI Connect is in part a response to the increasing number of conversations with customers across all verticals about generative AI (GenAI).
"The conversation and the air in the room has really been sucked up with discussions around GPUs, compute and power and [customers] are missing a key ingredient in activating that AI ecosystem, and that's the role the network plays in delivering AI workloads at scale," said Lawrence.
Network infrastructure will also play a key role in supporting AI initiatives such as Stargate, said Lawrence. Stargate is the $500 billion AI plan announced this month that includes backing from SoftBank, Oracle, MGX, OpenAI and more.
Shift to AI inferencing
Having sufficient support from network infrastructure will be key as AI moves from the cloud, to the edge, to everywhere – where AI is embedded "from devices to the edge to the core to the cloud," said Lawrence. The network will also be a vital component as industry focus on AI shifts from building large language models (LLMs) to also developing inferencing models, said Lawrence.
In its press release touting its AI efforts, Verizon quoted McKinsey & Company as a source projecting that 60-70% of AI workloads will shift to real-time inference by 2030. Inferencing is how AI applies what it learns and provides insights about new data.
Susan Welsh de Grimaldo, Sr. Director Analyst for Gartner, told Light Reading that the early wave of AI development focused more on training and is now shifting to inferencing.
"We're moving more toward the inferencing piece which will start to be more decentralized. The cost is going up and as enterprises really adopt AI applications they're seeing that they need to manage their FinOps (financial operations)," said Welsh de Grimaldo.
Enterprises also need to balance privacy concerns and might want the data closer to their operations to address latency issues. This presents an opportunity for edge AI use cases, but "we also still need the big fiber transport and metro fiber as well for the big data centers and connectivity there," said Welsh de Grimaldo.
Service providers have a big opportunity to leverage current network assets and continue to build on them for future AI requirements, said Welsh de Grimaldo. Operators should focus on building out key partnerships that expand their role in the AI ecosystem and position them to meet the AI needs of customers, she added.
"At this stage of the AI growth and adoption, it's not so much about winning," said Welsh de Grimaldo. "It's really more about learning and partnering to develop some value propositions that will ultimately generate some more revenue at scale for them. And they'll generate revenue because they'll deliver value to the customers and partners."
AT&T and Lumen are also getting in on AI infrastructure
AT&T and Lumen are also making significant moves to support the AI boom. AT&T recently announced a structured sale-leaseback deal for some of its underutilized central office facilities with private real estate development firm Reign Capital, reported Light Reading's Mike Dano. While the company didn't mention AI in the release, AT&T officials in the past have discussed the possibility that AT&T central offices could host AI capabilities.
Last November, Lumen Technologies reported that it has raked in $8 billion in private custom fiber (PCF) agreements with webscalers including Microsoft, Amazon, Google and Meta.
You May Also Like