BOSTON – The consensus among speakers and attendees at a recent AWS-hosted event was that scaling AI requires more than new models and better data infrastructure. Enterprise organizations must also expand network capacity to connect their data centers, cloud providers and employees in the field.
As businesses move forward with AI, the network is increasingly important for processing data closer to where it's generated – especially at the edge. This resonated with leaders at organizations of every scale and level of AI adoption at Inside AWS Boston on Nov. 4 -- a 3-year-old robotics company with a business model that relies on AI, an 80-year-old pro sports team with hundreds of employees that only deploys AI selectively, and a global telecom giant undergoing digital transformation.
"We're a 150-year-old company that's been through a lot of M&A. You can imagine what our data infrastructure looks like. We have a lot of legacy application infrastructure," said Luke Corbin, assistant vice president, digital transformation and artificial intelligence at AT&T, during an event presentation. "How do you build all of that infrastructure and integration that's going to allow you to build AI even faster? What is your actual connectivity backbone?"
Gartner's 2026 Market Guide for Edge Computing Platforms predicts that by 2028, more than two-thirds of enterprise-managed data will be created and processed outside the data center or cloud. According to IDC FutureScape: Worldwide IT Industry 2026 Predictions, 50% of all enterprise AI inference workloads will run on endpoints or edge nodes by 2030.
"When you're moving as much data as we are, and you've built your business to be dependent on it, you have to really think about what's moving between data centers and cloud providers, and [whether] you have the right high-capacity network to manage that," Corbin said.
AI demands edge network upgrades
In AT&T's case, workers in the field are becoming more dependent on AI agents, making reliable edge network connectivity essential to performing their jobs, Corbin said.
"When you have a field tech or a salesperson who's now dependent on this agentic layer for omnipotent and omniscient enablement of their job, they need to be able to connect, whether that's over mobility or over satellite [through our partnership with you all]," he said.
AT&T Business disclosed plans on Sept. 8 to integrate Amazon's Leo satellite network for remote areas into its enterprise and public service connectivity architecture.
"The whole thing becomes so critical. It's not like you missed an email. You can't actually do your job," Corbin said.
The whole thing becomes so critical. It's not like you missed an email. You can't actually do your job.
Luke Corbin
Assistant vice president, digital transformation and artificial intelligence, AT&T
At Luminous Robotics, the concern is the same -- the network is instrumental for workers at the edge to complete their jobs. However, the workers in question are fleets of autonomous robots that deploy solar energy infrastructure alongside humans.
The company's LUMI robots use vision-language-action models for edge inference, enabling them to operate without constantly communicating with the cloud.
Still, Luminous Robotics eventually sends all data back to AWS for processing. Depending on the purpose of the transmission, latency requirements vary, said Jay Wong, Luminous Robotics' CEO, in an interview with TechTarget.
"Some things are much more sensitive in terms of performance. Those get pushed to the cloud [several times per second]," Wong said. "We're bandwidth constrained. We're in the middle of nowhere typically. These [deployments] aren't like building a solar power plant in the middle of Boston."
We're bandwidth constrained. We're in the middle of nowhere typically. These [deployments] aren't like building a solar power plant in the middle of Boston.
Jay Wong
CEO, Lunimous Robotics
Luminous uses a combination of LTE and Starlink network infrastructure to connect its robots to the cloud. Anything that requires a higher transmission frequency – such as error logs or safety information – uses a high-speed LTE wireless network. Luminous reserves Starlink for workloads that require less frequent communication with AWS data centers and higher bandwidth.
Luminous also batch-uploads training data from robots on the ground to fine-tune its AI models in the cloud using AWS Bedrock, then sends updates to the robots over the air when needed. But Wong said the company generally avoids updating during active deployments.
"It's typically when there's something problematic, we would try to do it overnight, or [if] a release has a ton of great features, then we'll do it once every month or so," Wong said.
As Luminous grows, improving network bandwidth capacity will be top of mind, he said.
"We just announced two or three weeks ago [that] we're sizing up our manufacturing facility to produce hundreds of these robots," Wong said. "Right now, we're in an 11,000-square-foot facility, but there is much more manufacturing we'll have to do. As we [scale], I think some of these networking bandwidth constraints are going to be more front and center for us."
Wong wasn't sure whether Luminous Robotics would partner with Amazon to expand its networking infrastructure in the future, but was open to exploring Amazon Leo.
"We're fairly agnostic in terms of that requirement for our robots," he said.
Celtics expand network security
The Boston Celtics are also growing. The business is on a "tremendous ramp-up, " and the network infrastructure has had to scale to match that, said Jay Wessland, the team's CTO, in an interview with TechTarget. He cited an increased volume of data generated both on and off the court as a catalyst for expanding its network infrastructure and monitoring capabilities.
"Every game is on television five different ways, and every stat can be delivered to you five different ways," Wessland said.
"We have a significant bunch of cyber tools at the network level and at the user level that feed into our third-party security partner," said Wessland.
Among those tools, the Celtics use ReliaQuest's GreyMatter to filter security data using AI and to forward alerts to a human for review as appropriate.
"Sometimes it just auto-generates [information] about packets or things that are happening at the network edge," Wessland said. "We see those alerts and our third party sees [them], so it's a significant amount of work at the network level."
Ben Lutkevich is an award winning writer for TechTarget's infrastructure site.
