Verizon touts $1B dark fiber deal for Google data centers as first of many

Verizon has decided that the best way to monetize a century of buried cable is to rent it to the machines. The telecom disclosed a deal worth more than $1 billion to lease dark fiber to Google for data center interconnection, framed as the opening move of a broader "AI Connect" initiative. CEO Dan Schulman expects additional AI-related agreements before year-end, potentially worth "multiple billions of dollars" over several years.

The mechanics are straightforward. Dark fiber is unused optical cable already in the ground. Activating it for hyperscaler backhaul carries minimal marginal cost and converts stranded infrastructure into recurring revenue. Meanwhile, Verizon is stripping copper from retired central offices and repurposing the real estate as compact data centers aimed at AI inference workloads. The thesis: latency-sensitive compute—robotics, remote surgery, autonomous driving—will increasingly live at the edge rather than inside a single hyperscale campus.

The strategic subtext is a quiet admission that consumer wireless growth has plateaued. Telecom carriers built out fiber networks for residential broadband and 5G backhaul, only to discover that the most lucrative traffic pattern is now machine-to-machine. By positioning itself as the connective tissue between hyperscale training clusters and distributed inference nodes, Verizon is attempting to climb the value chain without building its own foundation models.

The risks are familiar. Hyperscalers typically prefer to own their physical layer, and Google has its own fiber assets. Multi-billion-dollar projections from a single earnings call are not contracts. But the directional signal is clear: the AI buildout is creating a secondary market for telecom infrastructure that did not exist eighteen months ago, and incumbent carriers intend to capture it before specialized competitors do.