American Tower delays edge computing aspirations

According to the incoming CEO of American Tower, demand for edge computing hasn't materialized as the company initially expected.

"There's nothing in our multiyear guide for edge," said Steven Vondran during a recent investor event, according to Seeking Alpha. "I don't know exactly when it's going to be there."

That represents a delay from American Tower's initial expectations in 2021, when the company first announced its $10.1 billion plan to buy CoreSite Realty, one of the nation's biggest operators of data centers. At the time, American Tower estimated that the total addressable market for the "mobile edge" would hit $1 billion by 2026 and the total addressable market for the "metro edge" would reach $2 billion.

"It has slowed down a bit," Vondran said earlier this week, in discussing American Tower's edge computing timeline. He said American Tower is no longer holding to its initial edge computing guidance. "But I am convinced more than ever now that it's when, not if, that that's going to develop," he added.

Verizon also has pushed back its edge computing guidance. The company's networking chief, Joe Russo, recently said Verizon has not seen much demand for public edge computing services – such services are different from the private edge computing products that typically sit inside an enterprise's premises. But he said public edge computing "is absolutely something we're preparing for."

Other companies backing off from edge computing include Ericsson and Deutsche Telekom.

Testing the waters

Vondran is scheduled to take over as American Tower CEO early next year. He said this week that the company continues to work on edge computing tests and pilots worth "a few hundred thousand dollars."

Other companies are also testing the edge waters. Amazon Web Services (AWS) offered a detailed look at its work with Verizon to test super-low-latency 5G connections that allow remote musicians to play together in real time.

AWS's lengthy post on the topic explained that Verizon used the SyncStage platform from Open Sesame Media coupled with AWS public edge computing sites. Verizon has built almost two dozen such sites under an edge computing partnership with AWS first announced in 2019.

The demonstration also used Verizon's QoS API (quality-of-service application programming interface), which is based on CAMARA specifications from the Linux Foundation. Verizon is among a large and growing group of wireless network operators pursuing the opportunities around network APIs.

"Remote music collaboration use cases need audio communication performance consistency with ultra-low latency and low-jitter. This demo shows that the Verizon QoS API and multiple AWS Wavelength Zones enable Open Sesame Media's audio synchronization SyncStage software to deliver consistent 5G network performance for music collaborators in different geographical locations," according to AWS.

Big promises

Edge computing promises to dramatically improve not only the speed and performance of Internet connections, but also to upend the infrastructure of the world's global Internet network by placing computing services geographically closer to users. 

In some cases that could involve data center operators building sites in smaller cities. But it could ultimately involve computing resources placed at the base of cell towers and other remote locations.

However, there are a wide variety of edge computing implementations and architectures. For example, some implementations could involve private businesses putting computing resources on their corporate premises.

That's exactly what McDonald's announced with Google Cloud this week. In support of artificial intelligence (AI) and other technologies, the food vendor said it would "use edge computing from Google Cloud to power these new platforms, bringing information storage and high powered computing into individual restaurants."

Specifically, McDonald's said it would install Google Distributed Cloud at thousands of restaurants "so they can leverage both cloud-based software applications and their own software and AI solutions locally on-site, as needed."