The key to this deal is not just that "a telecom company has signed a major client," but that dark fiber is transforming from idle network resources into a part of AI capital expenditure. Google is purchasing a controllable physical transmission layer, which means that AI competition has extended from models and chips to interconnectivity between data centers.

In terms of capital, Verizon is not selling bandwidth packages, but long-term, low-churn infrastructure rentals. Schulman describes it as part of a major cycle in U.S. AI infrastructure, indicating that telecom operators are repackaging long-distance and metropolitan fiber assets into a "selling shovels" business for the AI era.

Historically, similar to the early explosion of cloud computing with data centers, submarine cables, and IDC expansion, such revenues are often underestimated until major clients begin to lock in capacity in bulk, at which point the market realizes that "connectivity" itself is a scarce resource. The difference this time is that the demand side comes from AI training and inference, rather than ordinary internet traffic.

Essentially, this represents a reconstruction of the industrial chain: AI is re-bundling computing power, storage, networking, and power supply, and the value of telecom companies is no longer just the last mile, but becoming upstream nodes of AI infrastructure. Whoever controls the physical connections will reap the spillover benefits of this round of capital expenditure.

ABAB News · Law of Cognition

The end of computing power expansion is network expansion.
Once connectivity becomes scarce, fiber turns into an asset.
Those who sell infrastructure will be the first to benefit from AI.