To many people, the concepts of global network infrastructure and bandwidth markets are difficult to grasp. But to those who follow this sector, it's one of the most fundamental building blocks of the global economy.
If you haven't heard of TeleGeography’s Transport Networks Research Service, it's one of our most popular research subscriptions. In short, this tool assesses the state of the global telecom capacity market—and we just finished updating it with tons of new data and analysis.
So now is the perfect time for us to answer the question: what's fueling the changes we see in the global bandwidth market?
Geopolitical Concerns
While geopolitical concerns have always played a role in determining which companies deploy long-haul networks where, several recent developments are reshaping network deployment trends.
The Red Sea is the focus of major problems. Even before the recent spate of rebel attacks on commercial shipping vessels, the Yemeni civil war created permitting headaches.
Cable laying vessels require permits to enter a country’s territorial waters, and when two different entities claim the same swath of sea, the situation becomes complicated.
Cable laying vessels require permits to enter a country’s territorial waters, and when two different entities claim the same swath of sea, the situation becomes complicated.
The nearly simultaneous cable faults on three cables off the coast of Yemen in late February have created a major challenge. Given the location of these faults, it is uncertain when maintenance vessels may be able to conduct repairs due to Houthi rebel attacks.
Subsea cable activity is also geopolitically challenging in the South China Sea. Cable builders find it increasingly difficult to receive Chinese permits for cable deployment in this region. Builders of the planned Apricot cable hope to avoid this problem by linking Japan to Singapore via the east side of the Philippines.
In addition, U.S. government opposition to direct China-to-U.S. cables has boosted the development of several cables from Southeast Asia to the U.S.
Artificial Intelligence (AI)
There's hardly a hotter topic in the network world right now than AI.
As the broader economy braces for the massive but unknown impact of this technological development, we have to ask what kind of effect AI will have on the transport network market.
The answer? It depends.
Model Training
If a given model is pulling data from all over the world, it will increase demand for long-haul bandwidth, but if data used is stored locally in the same campus or data center, it won't.
Data residency requirements will also factor into the use of long-haul networks for model training.
Training Clusters to Inference Clusters Data Transfer
If models are largely trained in more remote locations with cheaper and more abundant power sources, long-haul capacity will be needed to transfer data from training to inference locations.
The size and frequency of training model updates will also factor into this aspect of demand. Of course, if training needs to be located closer to inference zones, long haul capacity won't come into play as much.
Inference
The location of inference clusters will also be critical in determining the ultimate effect of AI on long-haul capacity requirements.
If inference is deployed in zones close to end users, it may not impact long-haul demand as much. If the model requires data pulls from multiple locations, though, it could increase long-haul demand.
These are just a few of the many factors to consider. As AI technology rapidly evolves, we expect that long-haul demand will grow, even if metro area demand increases more rapidly.
Private Cable Expansion
Google and, to a lesser degree, Meta are increasingly deploying their own subsea cables. This approach allows them to control the system design and landings, but also to move swiftly and avoid potential delays from working with partners.
Google is the owner of 15 active and planned private cables, with many more in development—particularly in the Pacific. Meanwhile, Meta is planning the Anjana trans-Atlantic cable, its first proprietary cable.
However, even on a private cable, content providers are not the only users. A portion of the fiber pairs are typically available for sale or swaps.
In the case of swaps, a content provider will swap fiber pairs in exchange for fiber pairs on other cables, for landing rights in a country, or for terrestrial backhaul within a country.
Content providers are not just swapping fiber pairs, but also selling them.
Content providers are not just swapping fiber pairs, but also selling them. Google is also selling IRUs for whole and partial fiber pairs on the company's private cables. Presumably, Meta would make fiber pairs available for sale on Anjana.
Google's private cable investment is spread globally, but especially focused in the Pacific, where the company's Pacific Connect Initiative is crisscrossing the ocean with multiple cables. Notably, Google is receiving some financial support from the U.S. and Australian governments for these projects, largely to enhance connectivity to islands in the South Pacific.
Completely private content provider cables are unlikely to become the dominant model on every route.
In some countries, they may not possess the legal or regulatory authority to land cables themselves and may need to rely on carriers or specialist companies who can operate a cable on their behalf. Also, few content providers currently have enough bandwidth demand to justify investing alone in a new cable.
Download the new Transport Networks Executive Summary to keep reading our latest analysis.
Jon Hjembo
Senior Research Manager Jonathan Hjembo joined TeleGeography in 2009 and heads the company’s data center research, tracking capacity development and pricing trends in key global markets. He also specializes in research on international transport and internet infrastructure development, with a particular focus on Eastern Europe, and he maintains the dataset for TeleGeography’s website, internetexchangemap.com.