Fiber infrastructure provider Zayo has partnered with Nvidia to build out network capacity aimed at supporting AI factories, according to Mobile World Live.
The stated goal is to close a gap between compute demand and bandwidth — in plain terms, the chips that train and run AI models have gotten far faster than the fiber links that carry data between the buildings housing them. Mobile World Live reports the partnership is designed to address exactly that mismatch.
Separately, Fierce Network reports that Zayo has confirmed Nvidia as its new "monster" customer, a detail that turns a vendor partnership into something more pointed: Nvidia is not just a technology collaborator here but a major buyer of Zayo's capacity. Light Reading, covering the same announcement, frames it as Zayo teaming with Nvidia to scale network capacity for AI factories.
"AI factories" is Nvidia's term for the very large data centers built specifically to produce AI output at industrial scale. These sites increasingly cannot fit on a single campus, which means the connections between them — long-haul fiber, metro routes, the unglamorous plumbing of the internet — become a hard limit on how much AI a company can actually run.
The available reporting does not disclose financial terms, capacity figures, or a timeline.
Why it matters: the AI boom is usually told as a story about chips, but a GPU is only as useful as the data you can feed it — and deals like this one show the bottleneck quietly shifting from processors to the fiber connecting them.