There are not many businesses that look safer because of AI. For most companies, AI is a new source of risk. It can compress margins, automate part of the value chain, lower barriers to entry, or turn today’s product into tomorrow’s commodity.The obvious winners are easy to name and mostly expensive. Semiconductors, data centers, power equipment, grid gear, cooling, and transformers all trade at prices that assume years of growth.There is another layer of infrastructure that gets far less attention. It does not train models or sit inside a GPU cluster. But without it, AI never reaches the user.Every prompt, every generated image, every enterprise agent, and every connected car has to cross a communications network. AI is created in the data center, but it becomes useful only once it travels to people, homes, devices, and businesses. The last mile is part of the stack.A recent traffic forecast shows the direction:Source: OmdiaAs AI moves from occasional chatbot queries to multimodal apps, always-on assistants, agents, and machine-to-machine workflows, the load does not stay in the data center. It pushes outward, into the access network.The company below owns one of the largest last-mile communications toll roads in the United States. More intelligence, more media, and more devices moving through the cloud all have to ride across networks like this one.The market does not price it like an AI beneficiary. It does not price it like ordinary infrastructure either. At today’s price, the stock trades at roughly:3.9x last-twelve-month earnings3.3x next-twelve-month earnings5.7x TEV / EBITDAThose are not AI multiples. Those are “something is broken” multiples.And maybe something is broken. The company carries debt. Competition is intensifying. It is spending heavily to defend and upgrade the network. The old story of a clean compounder is gone.But the stock does not need a premium multiple to work. It does not need the business to become wonderful. It needs the asset to stay good enough. If cash flows stabilize, the capex cycle rolls over, the pending merger integrates cleanly, leverage moves toward target, and the company spends the rest on buybacks, the per-share math changes fast.In a reasonable upside case, I think the stock can return 5x over five years.The setup exists because a hated infrastructure asset is priced as if it is dying, at the moment the world is about to push more traffic across its network than ever before.In the sections below, I will name the company, walk through the business model, explain why the market hates it, lay out the competitive threats, unpack the merger, model the debt and capex cycle, and show how buybacks could turn a low-growth asset into a high-return equity.