Artificial intelligence (AI) is scaling at extraordinary speed, reshaping industries, and redefining how data is created, processed, and consumed. Massive training clusters and increasingly ubiquitous inference systems suggest a future where intelligence is everywhere. Yet beneath this progress lies a less visible constraint: global WAN and backbone network connectivity.
If not addressed, the backbone network capacity and connectivity, not compute, will become the limiting factor in AI’s growth.
Compute Scales Fast; Networks Do Not
Driven by high-performance GPUs, specialized accelerators, and distributed software frameworks, AI infrastructure has advanced rapidly. Organizations can deploy new compute capacity in months, sometimes weeks.
Network infrastructure operates on a different timeline.
Terrestrial fiber builds require permits, construction, and coordination across jurisdictions, often taking years before a single strand goes live. Subsea cable systems can take five years or more to design, finance, and deploy. Even upgrades to existing systems are bound by physical and economic limits. As a result, while compute scales elastically, connectivity expands incrementally.
This dynamic creates a growing imbalance. AI systems are becoming more distributed and data-intensive, while the networks connecting them struggle to keep pace.
This story is part of a paid subscription. Please subscribe for immediate access.
ABOUT THE AUTHOR
Dr. Jason Black leads the teams who are driving the delivery of NVIDIA’s end-to-end data center-related software and infrastructure that power today’s most comprehensive cloud and edge data center deployments. Formerly Black was at Uber Technologies, where he was Head of Global Network Infrastructure covering data center, backbone, PoP, and cloud for the production and advanced technology group networks. Prior to his time at Uber Technologies, he was Senior Director of Network Engineering and Operations at Yahoo! He is a technology and business visionary with hands-on experience in growing multi-billion-dollar web scale companies as well as his previous start-ups.
