The world’s largest internet service providers are investing less in network infrastructure than they have in years, even as their revenues rise, according to more than a decade of data analyzed by Harvard scholar Susan Crawford and telecom analyst Mitchell Shapiro. This trend raises the question: could emerging technologies reverse it?
Shapiro’s report notes that Comcast’s capital expenditure (capex) to revenue ratio peaked at around 37 percent in 2001, driven largely by major acquisitions that required substantial network upgrades. Around 2000, many ISPs were shifting toward hybrid fiber-coaxial architectures—networks that combine optical fiber with existing copper wiring—to expand capacity without replacing every link end-to-end.
Part of the reason investment appears to have slowed is that much of today’s deployed cabling dates back to the early 2000s and has proven durable, reducing the immediate need for wholesale replacement. In some cases, older copper infrastructure has continued to surprise engineers: researchers at Bell Labs demonstrated multi-gigabit performance over legacy copper lines, showing that existing physical networks can sometimes be pushed well beyond their presumed limits.
Still, newer transmission approaches are under development that could transform how data is routed and distributed. One promising line of research, published by Dr. Milchberg and colleagues, explores using intense laser pulses to temporarily alter air’s optical properties and create transient pathways for light. Extremely powerful laser bursts can ionize and collapse air along a narrow trajectory called a filament. As the filament evolves, it leaves behind a low-density channel—a brief, hollow column of air with a lower refractive index than the surrounding atmosphere, effectively forming a reflective “tube.”
By arranging four laser-generated filaments in a square and firing a fifth beam through the center, researchers can trap and guide the central beam within a cage of low-density air. The physics here occur on ultra-fast timescales: while the filaments themselves vanish in about one trillionth of a second, the resulting low-density channel can persist for a few milliseconds. Although milliseconds may seem fleeting, they are more than enough time to transmit significant amounts of data. If perfected, this approach could enable temporary, cable-free “waveguides” for optical signals, opening the possibility of delivering high-bandwidth links to locations that are impractical or costly to wire physically.
The implications are wide-ranging. Beyond terrestrial applications—such as connecting remote communities or temporarily augmenting capacity at crowded events—these airborne waveguides could potentially be used to beam data to high-altitude platforms or even orbiting spacecraft. To date, Milchberg’s team has demonstrated the effect over a one-meter range with very low energy loss. Their next steps include scaling experiments to 50 meters and, if successful, developing practical systems for real-world deployment.
Whether such technologies will reshape ISP investment priorities remains uncertain. They could reduce the need for continuous physical upgrades in some contexts, or they could spur new types of infrastructure spending to deploy and maintain laser-based systems. Either way, the evolving landscape of transmission technologies suggests that the relationship between spending and performance will continue to change.
Do you think ISPs should be investing more in their infrastructure? Share your thoughts in the comments.