- Comcast and Optimum are testing whether AI can help them turn those networks into new sources of revenue
- Speaking at FutureNet North America 2026, executives from both companies described AI strategies that start with automating network operations and software development
- Elad Nafshi said Comcast sees three pieces to its AI strategy
FUTURENET NORTH AMERICA, NEW YORK CITY – Cable operators have spent years talking about squeezing more efficiency out of their networks. Now Comcast and Optimum are testing whether AI can help them turn those networks into new sources of revenue.
At FutureNet North America 2026, executives from Comcast and Optimum described AI strategies that start with automating network operations and software development but increasingly extend to creating new services on top of their existing infrastructure.
That shift comes as the traditional broadband business remains under pressure. Comcast reported that it lost 144,000 domestic residential broadband customers in the third quarter, although it added 368,000 wireless lines.
Optimum's latest reported quarter is Q2, with Q3 results scheduled for Nov. 6. In Q2, broadband net losses narrowed to about 40,000, while mobile delivered 50,000 net additions.
Comcast sees AI at the edge
Elad Nafshi, EVP and chief network officer at Comcast, said the operator sees three pieces to its AI strategy: using AI to operate and scale its network, making the network capable of supporting increasingly powerful AI applications, and finding ways to monetize the distributed compute infrastructure it has already built.
Nafshi said Comcast operates more than 50,000 servers across roughly 1,300 physical pods in 200 edge data centers. That infrastructure already powers the connectivity business, but Comcast is now asking what else it can run on it.
“How might we be able to monetize that?” Nafshi said, pointing to proximity to compute, lower latency and the ability to optimize workloads as potential advantages.
Comcast has already started putting additional applications on that infrastructure. Nafshi provided examples of a network-based firewall service for small businesses and a live linear cache designed to improve delivery of high-demand live events. He said those services are the beginning of a broader effort to turn the cable company’s network into a platform rather than a broadband pipe.
“We need to think and adapt to operate and learn and deliver as a software company, not as a connectivity company,” he said.
Optimum targets AI productivity
Optimum opened AI tools to employees across the organization without initially imposing token limits, according to Luciano Ramos, EVP and chief product and technology officer at the company. That experiment stretched well beyond developers to project managers, product analysts and QA testers, said Ramos.
The results have included non-technical employees building applications and teams automating forecasting, program management and reporting. Among software engineers, roughly 30% are producing five to six times more pull requests with AI, he said.
Optimum has also pushed automation deep into network operations. Ramos said areas including incident management, incident response and root-cause analysis are now heavily automated, with bots handling work and bringing humans into the loop when necessary.
The ROI metrics include faster time-to-production and faster software development rather than an AI transformation targets.
Ultimately, this morning’s kick off panel offered some clear insight into how two major cable operators are using AI in the face of tough competition. AI is beginning to split into two business cases: using automation to run networks and technology organizations faster and more cheaply; and figuring out whether distributed networks, edge compute and customer relationships can support services that customers will actually pay for.
Read more about AI on Fierce Network
Verizon taps Ericsson for rApps rollout
Lumen wants to use the network to help AI workloads chase power