Network providers face a familiar squeeze: traffic keeps climbing while legacy connectivity models become harder to differentiate. The industry’s response is a frontier model: a human-led, AI-operated organization where intelligence runs across employees, customers, operations and partner networks. Leaders still set strategy, but AI increasingly carries it out — summarizing information, surfacing next-best actions, coordinating software agents and enforcing guardrails at scale.
The stakes go beyond efficiency. As traditional connectivity becomes harder to differentiate, the companies creating value are the ones turning network and customer data into intelligence, not just using automation to lower costs. Connectivity alone is sold as a commodity; differentiation now lies in how data flowing across its network, for its own operations and for the enterprises it serves. That shifts the question from who has the biggest network to who can turn that network into an intelligence platform, which is why operators are pushing toward the frontier model rather than incremental automation.
Frontier organizations are equipping employees, from network engineers to enterprise sellers, with AI copilots and agents that cut manual work. They are using network and customer data to build personalized offers that improve conversion and retention. They are embedding intelligence directly into network operations, service delivery and go-to-market workflows. And they are using AI to prototype and launch new services faster, including industry-specific solutions and autonomous network capabilities.
Kevin Shatzkamer, CVP, Engineering - Telecom and Media Studio for Microsoft Frontier Company, says “Microsoft Frontier Company is not a product. It is engineers embedded directly inside customer environments, building and running AI alongside the people who own the work. The operating thesis is that forward deployed engineering is largely the right model, and that the real value sits in co-design, co-deployment, and continuous improvement of AI systems at scale. Every system is outcome-driven by design, because at this point a system that cannot deliver a measurable business result is not worth deploying.”
Kevin describes what he asks of his own team in a single line: Fall in love with our customer's problems and not our products.”
Growing beyond connectivity
Industry analysts see a $60 billion to $100 billion growth opportunity for AI in telecom over the next few years, driven by new services, smarter operations and AI-powered customer experience.
Some companies are already showing results. AT&T reports roughly a 5x return from AI across operations and finance. T-Mobile is turning about 1 trillion in daily transactions into actionable insights. Vodafone’s AI assistant, TOBi, now handles about 45 million customer interactions a month. Lumen is transforming its network into a digital platform that gives enterprises greater control over how, where, and when their data moves, a shift explored in detail below.
B2B AI offerings are projected to grow at a 65% compound annual growth rate as enterprises seek partners that combine connectivity, edge and cloud intelligence. McKinsey estimates hyper personalization alone can lift ARPU 5% to 15%, mostly through converged bundles and premium digital services.
Most companies aren’t built for that yet. Pricing, RFP responses and customer signals still move through siloed tools, spreadsheets and manual approvals, and frontline teams aren’t equipped to sell AI-based offerings. Closing that gap means investing in skills and workflows, not just platforms.
Microsoft’s answer is what it calls an IQ layer. Work IQ brings intelligence into daily workflows through tools like Microsoft Copilot; Fabric IQ unifies data across silos in a single analytics and governance layer; and Foundry IQ provides a secure environment for building and testing AI agents. Above them sits Agent 365, a control plane that gives AI agents their own identities, scoped permissions and auditable access to data.
The stakes are rising alongside the opportunity. Agentic AI across telecom and network services is projected to generate roughly $150 billion in value by 2030, but companies will only capture that value if they can secure their networks and data. Most companies in the sector have already reported significant breaches.
Kevin explains why Lumen’s strategy and Microsoft are totally complementary: “The partnership combines Microsoft's platform depth, engineering expertise, product knowledge, and cross-industry learnings with Lumen's AI-ready network, customer insight, operational expertise, and accountability for business outcomes. These complementary strengths create a build-with model that accelerates delivery while ensuring that resulting capabilities are grounded in real-world enterprise environments and can be sustained at scale.”
Lumen: making the network the ROI lever
Lumen, which bills itself as “the trusted network for AI,” is building a programmable network that customers control. It’s a shift beyond connectivity as a product.
Ryan Asdourian, executive vice president and chief strategy and marketing officer at Lumen, says that when customers can control where and how their data moves, the network stops being a commodity and becomes a strategic lever. “You suddenly have a different interest [from] customers,” he says.
Customers, he adds, “look at the features we're able to give them, and they see a digital platform that is able to help them achieve their business goals. That's a fundamentally different value proposition than what our industry has provided in the past.”
The strategy is showing up in the numbers: NaaS customer adoption up 22%, services sold up 29%, and active fabric ports up 34% quarter on quarter — double-digit gains Asdourian attributes to customers wanting more control over how their data moves.
“In our transformation, you can have the best tech in the world, but you also have to have the
culture to support people in selling and delivering AI enablement,” Asdourian says. “It’s really about how we ensure our people are ready to serve our customers in the way that they need, and our culture is the foundation of that.”
Lumen has deepened its partnership with Microsoft since mid-2024, modernizing its own operations on Azure while supplying network infrastructure for Microsoft’s AI build-out. Inside Lumen, Microsoft Copilot has moved from pilot to everyday use.
“AI didn't just create more traffic. What it did was it changed the way customers had to think about their infrastructure because the bottleneck had moved,” Asdourian observes.
On the network side, Azure Local lets Lumen bring enterprise data and telemetry closer to Azure for latency- and sovereignty-sensitive scenarios, training models near the edge and feeding insights back into Copilot and operational systems.
“Lumen's goal is to make sure we are not the bottleneck, but we are the enabler,” Asdourian explains.
Microsoft frames this shift as “Return on Intelligence,” using the network as an AI ROI lever across cost, speed and performance. At Lumen, Asdourian argues the network is the key lever in that equation.
Much of the world still runs on 10 gig circuits. Moving a single petabyte of data takes about 222 hours, or more than nine days. Over a 400-gig connection, it takes roughly six, a 40-fold reduction.
At about $2,000 an hour, a rented GPU cluster bills through the entire transfer before training can begin. Asdourian puts the difference at $431,000 per petabyte. “That was tolerable when it didn’t cost you $431,000 per petabyte of data,” he says. “Now it’s unacceptable.”
Asdourian says AI's biggest effect at Lumen isn't efficiency; it's the ability to deepen “customer obsession” by understanding customer needs and responding at scale.
“The real gains are the productivity for… every member of our company to achieve more, and that’s what we lean in on,” he concludes.
What’s next in the series
This story sets the stakes for frontier companies. Future installments go deeper. In December, Microsoft addresses unifying data for network-to-enterprise intelligence. In February, the series explores personalizing customer experience at scale. In June, it examines moving toward autonomous networks and AI-managed operations. Each piece will pair practitioner insight with concrete use cases and design patterns from operators making the shift.
To explore the full Microsoft Frontier Company framework, including detailed use cases, reference architectures and partner stories, visit https://www.fierce-network.com/activation-center/connect-future-ai.