Telecom has a capacity glut – so stop moaning about AI traffic

Telecommunications operators are currently managing a significant surplus of network capacity despite persistent industry warnings regarding the potential strain of artificial intelligence traffic. Recent market research suggests that the narrative of a looming capacity crunch may be overstated by technology vendors seeking to promote infrastructure upgrades. While the shift towards generative artificial intelligence necessitates high-bandwidth connections, current data shows that the global subsea and terrestrial fibre pipelines remain underutilised. This abundance of bandwidth indicates that existing investments in fibre-to-the-home and fifth-generation mobile networks are providing a substantial cushion against rising data demands.

Analysts point to the decoupling of traffic growth from capital expenditure as evidence that the era of desperate capacity expansion is ending. In previous decades, telecommunications firms were forced to spend heavily to keep pace with the explosion of video streaming and mobile data. However, the efficiency of modern spectral technologies and the maturity of core routing hardware have allowed capacity to outpace consumption. Even as large language models require massive datasets for training and inference, the actual pressure on the public internet backbone is reportedly less severe than the historical transition from standard to high-definition video.

The behaviour of major network operators further supports the conclusion that a capacity glut exists across many developed markets. Many service providers have slowed their annual investment cycles, choosing instead to focus on monetising existing assets rather than rushing into next-generation hardware deployments. This shift in strategy suggests that the industry is no longer fearful of being overwhelmed by data surges. Furthermore, the pricing for wholesale transit and international bandwidth continues to decline, which is a classic economic indicator of an oversupplied market.

The discourse surrounding the impact of artificial intelligence often focuses on the high-performance computing needs within data centres rather than the transmission networks connecting them. While intra-data centre traffic is indeed surging, the wide-area networks managed by traditional telcos are not experiencing the same level of stress. Experts suggest that the industry should shift its focus from mere capacity acquisition to the intelligent orchestration of existing resources. By optimising how data moves across the current surplus, operators can maintain service quality without committing to premature infrastructure rollouts.

Market participants are expected to continue monitoring traffic patterns as more consumer-facing AI applications are launched globally. The focus for the coming fiscal year will likely remain on enhancing the software layer of the network to better handle varied traffic types rather than laying new physical fibre. As the gap between available capacity and actual usage remains wide, the industry is positioned to absorb the next wave of technological innovation without significant financial strain. This stability will likely influence future procurement cycles and the long-term deployment strategies for advanced telecommunications services.