The increasing codependency and simultaneous growth of artificial intelligence (AI) and fiber optic networks can no longer be seen as separate investment cycles, and as AI moves beyond centralized training to distributed, real-time use cases, fiber will become the nervous system that enables intelligence and data to be moved, coordinated, and put into action, says the Fiber Broadband Association (FBA).
of Building a thinking economic nervous system In this paper, we investigated how the convergence of fiber broadband and artificial intelligence is reshaping economic infrastructure. It provided broad guidance on how fiber will shape future economic growth, including public safety, health care, manufacturing, grid modernization, national security, and local competitiveness.
This paper argues that the United States is currently experiencing two simultaneous infrastructure supercycles: national fiber broadband deployment and rapid build-out of AI computing infrastructure, and that these cycles are converging into a single ecosystem, with significant implications for network operators, infrastructure investors, data center developers, communities, and policy makers.
But the central proposition is straightforward. Fiber is no longer just an access technology. It is becoming the nervous system of the thought economy, the connective layer through which intelligence is generated, distributed, and acted upon on a large scale.
In fact, this study quantified the fiber infrastructure gaps that these convergence cycles create. FBA/RVA research estimates that the U.S. may need to nearly double fiber route miles from 95,000 miles to 187,000 miles and increase total fiber miles from 159 million miles to 373 million miles by 2029 to support AI and data center growth. FBA added that each new hyperscale data center requires an average of 135 new route miles of connectivity.
The study cites findings from the International Energy Agency (IEA) that total data center power consumption will double by 2030, and power use from AI-focused data centers will triple. Additionally, the U.S. Department of Energy and Lawrence Berkeley National Laboratory predict that data center load growth in the U.S. has already tripled over the past decade and is expected to double or triple by 2028.
The fiber deployment supercycle is already of historic proportions, with U.S. fiber providers installing a record 11.8 million homes in 2025, bringing total U.S. fiber installations to 98.4 million, including multiple installations and 84.6 million unique residences, according to data from the Fiber Broadband Association and RVA Market Research. This equates to 11% annual growth.
According to FBA, two-fifths of fiber broadband deployments in 2025 will be from non-tier 1 operators. This reflects a growing adoption ecosystem that includes electric cooperatives, municipalities, private equity-backed platforms, and competitive overbuilders.
The concurrent AI infrastructure supercycle is similarly unprecedented. According to the report, Amazon reported $131.8 billion in capital expenditures in 2025, primarily in AI infrastructure, AWS data centers, networking equipment, and custom AI chips. Alphabet announced plans to raise $80 billion in new equity, in addition to expected $180 billion to $190 billion in capital spending in 2026, specifically to fund expansion of its AI computing infrastructure.
Meta reported 2025 capital expenditures of $72.22 billion. Microsoft’s 2025 annual report lists more than 400 data centers across 70 regions, adding more than 2 gigawatts of new capacity in fiscal year 2025.
OpenAI’s Stargate initiative announces a $500 billion investment in U.S. AI infrastructure over four years, starting with $100 billion in immediate deployment. Our partnership with Oracle in July 2025 added 4.5 gigawatts of data center capacity. And with the September 2025 expansion by SoftBank, OpenAI, and Oracle, Stargate’s planned capacity will reach nearly 7 gigawatts and total planned investment will reach more than $400 billion over three years.
