“By proactively designing for multiple unpredictable failures, we help customers maintain business continuity even during network disruptions,” AWS director of global connectivity infrastructure development (GCID), David Selby, wrote in a blog post.

As a result, this philosophy helps shape how the company plans, builds and protects the global infrastructure which keeps the internet running.

AWS has built more than nine million kilometres of this network infrastructure, enough to stretch from Earth to the Moon and back more than 11 times.

Currently, the company operates a global infrastructure spanning 38 regions and 120 availability zones, all connected through redundant, high-capacity routes.

“We take a measured approach to building and expanding this infrastructure. Our network expansion follows careful analysis of customer needs and usage patterns,” Selby said.

“This ensures we build the right capacity in the right places at the right time, while preparing for our customers’ future needs.”

Fastnet: AWS’s new transatlantic subsea cable

“The AWS network operates across multiple diverse routes in every major subsea corridor worldwide,” he said.

This comes as the technology giant just announced Fastnet, a new transatlantic subsea fibre-optic cable system linking Maryland, USA and County Cork, Ireland.

The new cable will feature advanced optical switching technology, enabling seamless rerouting of data as future network needs evolve.

With a design capacity exceeding 320 terabits per second (Tbps), Fastnet will become part of AWS’s vast global network. For example, the system delivers over 320 terabits per second, enough to stream 12.5 million HD films simultaneously.

Over the past decade, AWS expanded its subsea backbone through direct investments and partnerships, working with groups such as the European Subsea Cable Association (ESCA) and the International Cable Protection Committee (ICPC) to continuously assess potential risks to its subsea cables after installation.

Meanwhile, it also works with local fishing communities to prevent accidental damage, advocating for greater industry investment in marine repair capabilities.

However, in areas at higher risk from earthquakes or shipping activity, Selby revealed its network automatically reroutes traffic to “help maintain connection stability” thanks to its extensive redundancy.

Strategic data-centre placement

While AWS’ subsea system connects continents, the technology giant’s land-based network links its data centres and customers.

“Each new data centre brings a new point of connection to the broader AWS network, regularly expanding our infrastructure by thousands of miles of fibre to bring cloud resources closer to where customers operate,” Selby noted.

Additionally, the company is scaling its data centres as a result of growing demand for AI and computing power.

“AWS designs a cohesive global network where every new connection strengthens the entire system’s resilience,” he said.

The company also revealed it has upgraded much of its infrastructure to 400 Gigabit Ethernet (GbE) technology, four times faster than before.

Meanwhile, the company is also developing new fibre technologies, including hollow-core fibre, which the company claims can improve transmission latency by 30% and multi-core fibre and increase data centre capacity.

“AWS invests heavily in both the hardware and software that runs our optical transceivers, which are crucial for high-speed data transfer,” Selby stated.

“These devices convert electrical signals to light for transmission over fibre optic cables and vice versa, connecting devices within our data centres and establishing high-bandwidth, low-latency connections to our customers’ networks.”

Scaling networks for AI workloads

The company also claims the increasing popularity of AI and ML workloads continue to create greater demand for its services, adding significantly more traffic across its network.

This comes as AWS unveiled its $38bn cloud computing deal with OpenAI, providing the AI giant with Amazon EC2 UltraServers, featuring “hundreds of thousands of chips” and the ability to scale to tens of millions of CPUs for its advanced generative AI workloads.

However, Selby also noted the “importance of streamlining the process for building new capacity to meet future needs.”

This comes as the technology giant just announced Fastnet, a new transatlantic subsea fibre-optic cable system linking Maryland, USA and County Cork, Ireland.

The move comes after CEO Andy Jassy warned that AI could replace some white-collar roles.

“While this will include reducing in some areas and hiring in others, it will mean an overall reduction in our corporate workforce of approximately 14,000 roles,” Jassy wrote.

“We’re working hard to support everyone whose role is impacted, including offering most employees 90 days to look for a new role internally (the timing will vary some based on local laws), and our recruiting teams will prioritise internal candidates to help as many people as possible find new roles within Amazon.”

RELATED STORIES

Subsea cable linking USA and Ireland unveiled by AWS

AWS, OpenAI sign $38bn cloud computing deal

AWS activates Project Rainier, one of the largest AI compute clusters