AWS Will Launch Sta'O'Nuk Subsea Cable in 2029
The 420 Tbps system will route traffic through Washington to mitigate transpacific infrastructure risk.
Updated on Oct. 8, 2026 in Data Centers
Live Poll
Is now a good time to significantly increase investment in national internet and AI infrastructure?
Amazon Web Services has announced the Sta'O'Nuk project, a high-capacity subsea cable system expected to go live in 2029. The network is designed to improve transpacific connectivity for AI training and real-time inference.
Why it matters
The system addresses growing concentration risk in subsea infrastructure, which current development trends show is becoming increasingly vulnerable to physical damage. By utilizing new landing points, AWS aims to ensure consistent performance for distributed AI workloads.
The Sta'O'Nuk system provides 420 Tbps of bandwidth across 20 fiber pairs. This capacity expansion serves as a response to the multi-year extension in subsea construction timelines, which have increased from 3 years to 5-7 years.
The players
Amazon Web Services
A global cloud provider specializing in scalable compute, storage, and networking infrastructure for distributed AI and enterprise applications.
The details
The system creates a new network path by landing in Hogan's Corner, Washington, bypassing traditional hubs in California and Oregon. It utilizes a combination of multilayered encryption, diverse routing, and continuous monitoring to maintain resilience. These methods are designed to protect against physical infrastructure threats, such as those observed in 2024 when four cables were severed off the Ivory Coast.
Timeline
2024: Four subsea cables were damaged near the Ivory Coast.
September 2026: AWS published a blog post regarding the project.
2029: The Sta'O'Nuk system is expected to become operational.
The Tech Race
The project follows a pattern set by the 2024 subsea cable damage off the Ivory Coast by prioritizing geographically diverse landing points to reduce systemic network failure risk. This effort signals a move toward more resilient, distributed network architectures required for high-stakes AI training.
This project will initially benefit enterprises and developers requiring reliable, low-latency connectivity for AI workloads across the Pacific. Users should expect performance improvements by 2029 as the network adds 420 Tbps of dedicated capacity to the transpacific route.
The takeaway
The move reflects an industry-wide shift toward hardening cloud infrastructure against physical and geographic bottlenecks. Watch for the 2029 operational milestone to see if the increased development cycle of 5-7 years remains the industry standard for transpacific projects.
Further reading
For more on the current state of infrastructure, see our coverage of Data Centers.
Source note: This article includes information reported by RCR Wireless News.
Live Poll
Is now a good time to significantly increase investment in national internet and AI infrastructure?
