- The system connects distant facilities to run massive training tasks continuously
- High-speed fiber keeps GPUs active by avoiding slow data bottlenecks
- Two-tier chip density increases computing power while lowering inter-rack latency
Microsoft has unveiled its first AI superfactory, connecting large AI data centers in Wisconsin and Atlanta through a dedicated fiber network designed for high-speed movement of training data.
The design places chips close together across two floors to increase density and reduce latency.
It also uses extensive cabling and fluid systems arranged to handle the weight and heat produced by large clusters of hardware.
A network built for model training on a large scale
In a blog post, Microsoft said this configuration will support massive AI workloads that differ from the smaller, more isolated tasks common in cloud environments.
“This is about building a distributed network that can act as a virtual supercomputer to tackle the world’s biggest challenges,” said Alistair Speirs, Microsoft’s general manager focused on Azure infrastructure.
“The reason we call this an AI superfactory is that it’s running a complex job across millions of pieces of hardware…it’s not just a single site training an AI model, it’s a network of sites supporting that one job.”
The AI WAN system moves information across thousands of miles using dedicated fiber, partly newly built and partly recycled from previous acquisitions.
Network protocols and architecture have been tweaked to shorten paths and keep data moving with minimal delay.
Microsoft claims this allows remote sites to collaborate on the same model training process in near real-time, with each location contributing its share of the calculations.
The focus is on maintaining continuous activity across a large number of GPUs so that no device pauses while waiting for results from somewhere else.
“Leading in AI isn’t just about adding more GPUs — it’s about building the infrastructure that makes them work together as one system,” said Scott Guthrie, Microsoft’s executive vice president of Cloud + AI.
Microsoft uses the Fairwater layout to support the high-throughput rack systems, including Nvidia GB200 NVL72 units designed to scale to very large clusters of Blackwell GPUs.
The company pairs this hardware with liquid cooling systems that send heated liquid outside the building and return it at lower temperatures.
Microsoft states that the operational cooling uses almost no new water, except for periodic replacement when necessary for chemistry control.
The Atlanta site mirrors the Wisconsin layout and provides a consistent architecture across multiple regions as more facilities come online.
“To make improvements in AI’s capabilities, you need bigger and bigger infrastructure to train it,” said Mark Russinovich, CTO, Deputy CISO and Technical Fellow, Microsoft Azure.
“The amount of infrastructure required now to train these models is not just one data center, not two, but multiples of that.”
The company is positioning these sites as purpose-built for training advanced AI tools, citing increasing parameter counts and larger training datasets as key drivers driving the expansion.
The facilities incorporate exabytes of storage and millions of CPU cores to support tasks around the primary training workflows.
Microsoft suggests this scale is necessary for partners like OpenAI and its own AI Superintelligence Team to continue model development.
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews and opinions in your feeds. Be sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, video unboxings, and get regular updates from us on WhatsApp also.
