Without addressing the energy, permitting, and connectivity bottlenecks, AI’s promised productivity gains will stall, reshaping investment strategies and policy priorities across the telecom sector.
The TeleGeography episode examines how the AI revolution is being constrained not by silicon chips but by the physical layers that move data—subsea cables, terrestrial fiber, and the power grid. Host Greg Bryan interviews Luis Colasante of Colt Technology Services, who argues that AI‑driven data centers now consume two to three times the electricity of traditional cloud facilities, making energy availability the ultimate gatekeeper for digital expansion. Colasante highlights three intertwined bottlenecks: insufficient grid capacity, a shortage of high‑speed fiber routes, and, critically, the permitting process that can stall projects for years. He cites real‑world examples, from a five‑year grid‑connection timeline for a new semiconductor plant in Syracuse to European fiber roll‑out timelines of two to three years, contrasted with hyperscalers’ 12‑to‑18‑month deployment expectations. The discussion also covers the “death of the toll model,” where raw bandwidth has become a deflationary commodity, prompting a shift toward Network‑as‑a‑Service offerings. The conversation draws a parallel to the late‑1990s telecom boom, noting that the earlier collapse was driven by speculative debt rather than technology failure. Today, hyperscalers control much of the subsea cable ecosystem—exemplified by France’s acquisition of Alcatel Submarine Networks—turning infrastructure into a strategic national‑security asset. Luis warns that this concentration creates systemic risk, especially when private‑equity investors demand rapid returns while infrastructure projects require long‑term capital commitments. For stakeholders, the takeaway is clear: sustainable AI growth hinges on coordinated investment in energy, permits, and connectivity, alongside regulatory frameworks that balance sovereign security concerns with market openness. Companies that can navigate these constraints and offer intelligent service layers will capture the next wave of value in the AI‑enabled economy.
Comments
Want to join the conversation?
Loading comments...