AI-rewritten: This is a summary of an article from Tom’s Hardware, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Xeal, an EV charging company, plans to install over 100,000 Nvidia GPUs across its existing U.S. network using a system called Latient Pods. The firm aims to create the world’s first edge inference compute network by utilizing idle electrical capacity at its more than 1,600 locations. Each pod contains up to 48 Nvidia Hopper or Blackwell Ultra GPUs and is designed to look like a standard broadband fiber cabinet found on streets.

Nikhil Bharadwaj, co-founder and CEO of Xeal, stated that these pods can deliver inference compute immediately while new data centers are still under construction. He noted that the system taps into permitted, grid-connected energy that currently sits unused, requiring no water hookups. The company claims this distributed network will provide ultra-low latency processing within metropolitan areas, with sub-20ms speeds compared to distant data centers.

Each Latient Pod features self-contained cooling and operates quietly at less than 65 decibels. Xeal says the units are built to withstand outdoor conditions in a NEMA 4 enclosure and can be installed and running in hours. The company reports that these pods offer fixed rental revenue for site owners, covering all power bills as a tenant pass-through while potentially adding up to $1 million in property value.

The first Latient Pod is scheduled to go online by the end of the year. Xeal emphasizes that its "Metro Edge" approach avoids the grid interconnect challenges often faced by new compute projects. However, security remains a concern; an Nvidia representative praised the innovation for better using existing infrastructure, while analysts note that these cabinets could be worth over $2 million each due to their high-value GPUs.