Featured Story
AT&T struggles to defend open cloudiness of Ericsson deal
More than a year into the Ericsson-led rollout, there is very little evidence AT&T's radio access network is as multivendor and virtualized as the telco makes out.
Verizon plans to sell the power, space and cooling necessary to run AI computing applications. Separately, AT&T is offering up its underutilized central office facilities through a real estate developer.
Both Verizon and AT&T are hoping to host AI systems on real estate they no longer need for their networking operations. The move essentially sets up the two companies to cash in on rising demands for AI data centers at the edge.
Verizon on Friday outlined the contours of the business opportunity it sees in the sector.
"If you think about where we are on generative AI today, it's where large language modules are trained at large data centers and that require enormous capacities. Over time, that will, of course, come much closer to the edge of the network," Verizon CEO Hans Vestberg explained on the operator's quarterly conference call.
Kyle Malady, head of Verizon Business and the executive leading the operator's AI efforts, offered more details: "Power, space and cooling are the currencies that are in demand right now, and we have all three," he said in discussing the AI sector. "As we look across our assets, take inventory and compare against other players in the market, we believe that we are in a leadership position when it comes to usable power and space. We have facilities across the United States that either have spare power, space and cooling, or can be retrofitted. As we sit here today, we have 2-10+ megawatts of usable power across many of our sites. ... In addition, we have between 100 and 200 acres of undeveloped land, some currently zoned for data center builds, and much of it in prime, data center-friendly areas."
Malady said Verizon would deploy Vultr's GPU-as-a-Service (GPUaaS) in its data centers in order to support the AI computing applications that require those kinds of high-performance graphical processing units (GPUs).
Malady added that Verizon sees a total addressable market (TAM) of $40 billion or more in this new area.
AT&T's real estate play
Along those same lines, AT&T on Friday announced a structured sale-leaseback deal for some of its underutilized central office facilities with private real estate development firm Reign Capital. AT&T said it won't need the facilities anymore as it shutters its aging copper network.
AT&T's transaction with Reign Capital covers 74 properties, located across the country, covering a total of 13 million square feet of space. AT&T said it will get $850 million initially alongside profit sharing from redevelopment efforts in the future.
AT&T didn't mention AI or data centers in its press release announcing the deal. An AT&T representative said it would be up to Reign Capital to propose uses for the real estate, which AT&T could then approve. However, AT&T officials in the past have discussed the possibility that AT&T central offices could host AI capabilities. For example, during AT&T's recent analyst event, CTO Jeremy Legg said that some of the company's central offices could be used for AI, but that the company would have to address the power requirements of those AI computing functions.
"Regional edge data centres are benefiting from the surge in demand for environments that can support AI model training," said analyst George Glanville of STL Partners in a release from the firm.
Broadly, STL Partners predicts the deployment of 1,800 network edge data centers by 2028. That figure represents a compound annual growth rate (CAGR) of 22%. Telecom companies are positioned to supply some of those data centers, STL Partners predicted.
The connectivity piece
Already telecom network operators are cashing in on AI, albeit through relatively standard data-transport deals. Specifically, they're selling dark and lit fiber connections to the hyperscale companies like Google and Meta that need to shunt AI data to and from their users.
Indeed, Verizon said it has already recorded $1 billion in those kinds of AI connectivity orders.
Other companies reporting similar deals include Lumen Technologies, which in November reported more than $8 billion in private custom fiber (PCF) agreements with the likes of Microsoft, Amazon, Google and Meta. And Lightpath said it booked nearly $110 million last year in AI transport agreements with hyperscalers.
Thus, selling the power, space, cooling and computing for AI would be the next logical step for telecom operators already transmitting AI data.
Indeed, that's the driver behind T-Mobile's recent agreement with Nvidia. The two companies see a future where T-Mobile's 6G network would be built atop basestations with Nvidia's AI-friendly GPU chipsets. Those chips could then be used to both run T-Mobile's network as well as AI applications for other companies.
We've been here before
To be clear, this is not new territory for telecom companies. Verizon and AT&T largely exited the data center industry roughly a decade ago. For example, Verizon sold around 29 data centers to Equinix in 2016 for $3.6 billion. Separately, AT&T sold its own data center colocation assets to Brookfield for $1.1 billion in 2019.
And just five years ago, both AT&T and Verizon hinted at broad ambitions to build edge computing facilities with the likes of Microsoft and Amazon Web Services (AWS). But those plans have not panned out either. That's because demand for edge computing services hasn't developed as many in the tech industry expected.
Will AI be different? Some believe so. That's because AI systems may soon shift from a "learning" model – currently running in big data centers – and toward an "inference" model that would rely on computing functions spread out across the country. For example, applications such as connected cars or factory floors may benefit from inference-based AI services, with edge computing facilities playing a major role in the speedy delivery of those services.
Verizon is clearly hoping that shift will happen.
"Initially, Vultr will deploy their GPU-as-a-Service infrastructure in one of our data centers and tap into our high capacity fiber network for distribution," said Malady, the Verizon executive. "We anticipate helping to broaden their reach and enable our mutual customers with AI training and inference capabilities at the edge over time. The bottom line is we are ready to help power the AI ecosystem."
Article updated January 24 to include commentary from AT&T.
You May Also Like