The increasing co-dependency and simultaneous growth of artificial intelligence (AI) and fibre optical networks can no longer be viewed as separate investment cycles, and as AI moves beyond centralised training and into distributed, real-time use cases, fibre will become the nervous system that allows intelligence and data to move, coordinate and translate into action, says the Fiber Broadband Association (FBA) in the US.
The Building the nervous system of a thinking economy paper studied how the convergence of fibre broadband and artificial intelligence is reshaping economic infrastructure. It offered a broad guide to how fibre will shape future economic growth, including public safety, healthcare, manufacturing, grid modernisation, national security and rural competitiveness.
The paper makes the case that the US is currently living through two simultaneous infrastructure supercycles – nationwide fibre broadband deployment and the rapid buildout of AI compute infrastructure – and that these cycles are converging into a single ecosystem with profound implications for network operators, infrastructure investors, datacentre developers, communities and policymakers.
Yet the central thesis is straightforward: fibre is no longer simply an access technology. It is becoming the nervous system of a thinking economy – the connective layer through which intelligence is generated, distributed and acted upon at scale.
Indeed, the study quantified the fibre infrastructure gap these converging cycles are creating. FBA/RVA research estimates the US may need to nearly double fibre route miles from 95,000 to 187,000, and increase total fibre miles from 159 million to 373 million, by 2029 to support AI and datacentre growth. FBA added that each new hyperscale datacentre requires an average of 135 new route miles of connectivity.
The study cites International Energy Agency (IEA) research projecting that total datacentre electricity consumption will double by 2030, while power use from AI-focused datacentres will triple. In addition, the US Department of Energy and Lawrence Berkeley National Laboratory estimate that US datacentre load growth has already tripled over the past decade and is projected to double or triple again by 2028.
According to the Fiber Broadband Association and RVA Market Research data, the fibre deployment supercycle is already historic in scale, with US fibre providers passing a record 11.8 million homes in 2025, bringing total US fibre passings to 98.4 million including multiple passings and 84.6 million unique homes. This represents 11% annual growth.
Two-fifths of the fibre broadband deployed in 2025 came from non-Tier 1 operators, reflecting, said the FBA, a broadening deployment ecosystem that now includes electric cooperatives, municipalities, private-equity-backed platforms and competitive overbuilders.
Happening simultaneously, the AI infrastructure supercycle is equally unprecedented. The report showed that Amazon reported $131.8bn in capital expenditures in 2025, largely for AI infrastructure, AWS datacentres, networking equipment and custom AI chips. Alphabet announced plans to raise $80bn in new equity specifically to fund AI compute infrastructure expansion, alongside projected 2026 capital expenditures of $180-190bn.
Meta reported $72.22bn in capital expenditures in 2025. Microsoft’s 2025 annual report describes more than 400 datacentres across 70 regions with more than two gigawatts of new capacity added in the 2025 fiscal year.
OpenAI’s Stargate initiative announced an intended $500bn, four-year US AI infrastructure investment beginning with $100bn deployed immediately; a July 2025 partnership with Oracle added 4.5 gigawatts of additional datacentre capacity; and a September 2025 expansion by SoftBank, OpenAI and Oracle brought Stargate’s planned capacity to nearly 7 gigawatts and total planned investment to more than $400bn over three years.
