That ambition is framed internally not so much as a pivot but as an escalation. McElfresh describes the decision as a reaffirmation of a role the company has played for decades: investing in the systems that made the modern internet possible in the first place.

He argues that many people still misunderstand what the internet actually is, seeing only the services on top of it rather than the networks that connect servers, devices, and the end users and locations. It is also the pathway through which people and businesses reach cloud computing resources, data centers, AI workloads, and large language models.

That framing matters because it speaks to a deeper strategic question facing AT&T, and really every incumbent in an era defined by AI exuberance. Is the company still best understood as a traditional telecom operator, or is it becoming something more like a connectivity platform or essential digital infrastructure? McElfresh’s answer is expansive. “All the above,” he says.

AT&T remains a telecom company, but one whose future depends on blending wireless and fiber into a single connectivity platform. More than that, McElfresh is making the case that AT&T wants to be the essential network layer of the AI economy. “Our vision is that AT&T becomes the highway upon which commerce and AI workloads traverse,” he tells Fortune.

It is a remark that captures both the scale of the opportunity and the risk. In the AI economy, attention tends to flow toward model makers, chip designers, hyperscale cloud companies, and app developers. But none of those businesses operate in a vacuum. They depend on resilient, high-capacity networks that can quickly and reliably move vast amounts of data. If McElfresh is right, the winners in the next phase of AI will include both those building the intelligence and those building the roads.

From AT&T’s perspective, the company has already been laying the groundwork for that thesis. McElfresh, who has spent some 30 years at AT&T, points to a strategic refocusing over the past several years on its core business of connectivity and convergence, with the company investing heavily in high-performing fiber. Today, he says, AT&T has connected 36 million locations with fiber and aims to reach more than 60 million endpoints, spanning homes, apartment buildings, schools, and government facilities.

The scale of the physical project is striking. Every month, McElfresh says, AT&T is physically installing more than 3,300 miles of fiber optic cable, roughly the distance from New York to Los Angeles. That analogy cuts through the abstraction that often surrounds conversations about AI. Somebody still has to build the underlying network. Permits have to be filed, trenches have to be dug, fiber has to be laid, workers have to be hired, and capital has to be allocated.

That is where the AI narrative shifts from theoretical to operational. McElfresh argues that AT&T is not only building the infrastructure required for an AI-driven economy, but also using AI internally to improve its ability to do so. The company uses AI to support technicians during home visits, assist engineers in determining where to place cell sites, and optimize fiber deployment. He says those tools are already improving productivity and enabling more accurate capital allocation decisions than the company was making just a few years ago.

For all the anxiety surrounding AI’s impact on jobs, McElfresh frames the technology first as a productivity engine that augments work already being done. AI is also becoming part of the machinery that helps AT&T execute its enormous spending plan more efficiently.

The company is already seeing AI begin to alter how data moves across its systems. During the pandemic, AT&T watched remote work redirect network use from downtown business districts to suburban neighborhoods in real time. Now, in the AI era, AT&T is seeing more data fed into the network rather than simply pulled from it, as connected cars and AI-driven software systems upload video, images, and other information to train models or receive their next instructions.

That emerging reality requires AT&T to transform itself alongside the network it is expanding. And that is where one of the interview’s central tensions comes into focus. Spending $250 billion over a decade suggests ambition. Still, it also raises a classic corporate problem: How can a company with the scale and infrastructure footprint of a utility maintain the speed and adaptability of a tech firm?

McElfresh says the challenge is not new, noting that AT&T’s longevity over roughly 150 years has depended on its repeated ability to reinvent itself. He traces that evolution from copper-based voice service to data over copper, then to mobility, and now to fiber-optics and gigabit-speed connectivity.

Perhaps the clearest sign of AT&T’s seriousness about its AI plans is the accounting. McElfresh says that beginning in 2026, AT&T will separate its financial reporting into an advanced connectivity profit-and-loss statement and a legacy one, giving investors a clearer view of where future services are growing and where older parts of the business are shrinking. That move suggests a company trying to force discipline around transformation. Separate reporting creates visibility for investors and gives management a scoreboard, making it harder to hide behind the inertia of a mixed legacy-and-growth portfolio.

There is a cultural dimension to all of this that McElfresh returns to repeatedly. He is a second-generation AT&T employee and has spent roughly three decades at the company, a fact that seems to inform both his reverence for the institution and his impatience with complacency. Asked what a company with this much history needs to guard against if it wants to define the next era rather than merely preserve the last one, he answers without hesitation.

“It’s being too comfortable with maintaining the status quo and managing the company that you have today, as opposed to stepping forward with a clear lens on the market, what kind of competition or technology is orbiting around your industry, and charging ahead on that to go build the company you want.”

That, he suggests, is the central challenge behind AT&T’s infrastructure push. The risk is not any single threat. Cyber threats, infrastructure failures, and geopolitical disruptions are real, but familiar. What keeps McElfresh up at night is the organization’s pace. “What AI is doing to us as a publicly-traded company, as an industry, or just in general, is the expectation that we can all move at a quicker clock speed, which requires a little level of discomfort.”

That brings him back to the $250 billion commitment, which he describes as a bet that AT&T, as both a telecom company and a connectivity platform, will remain central to the future it is helping build.