Xeal plans to deploy 100,000 Nvidia GPUs across 1,600 EV charging locations, feeding them with more than 200MW of permitted but idle electrical capacity. The company, which operates charging infrastructure in 500 US cities, announced the platform, called Laitent, this week.

The economics rest on a permitting gap. Charging sites receive permits for a maximum energy load but typically run below 10 percent of that ceiling, Xeal says. In theory, the remaining 90 percent can route straight to compute.

What does the platform actually consist of?

Laitent is a network of edge data centers built around Laitent Pods. Key specifications:

• Each pod occupies roughly one parking space.
• A pod houses up to 48 Hopper or Blackwell GPUs.
• Pods need no water connection.
• Dynamic power orchestration software can activate a single slice of one GPU or coordinate pods across several installations in one city.

Xeal targets AI inference workloads, positioning the pods where latency-sensitive demand sits: at grid-connected charging sites already inside the permitted envelope.

Co-founder and CEO Nikhil Bharadwaj framed the pitch around speed. "If you need inference compute today and can't wait for a new data center to come online, Laitent can accelerate your timeline from years to months," he said. "Laitent is the fastest way to bring new compute online, tapping already permitted and grid-connected, yet idle energy capacity, located right where it's needed for ultra-low latency, with no water hookup requirements."

The takeaway: Xeal is selling speed to power, not cheaper power.

Who is building it, and when?

Three partners cover the stack:

• Rafay Systems handles AI infrastructure orchestration.
• Spectrum Business supplies fiber connectivity.
• JVM Realty brings the pods online at its properties.

First deployment is expected before the end of the year.

Xeal also projects a financial upside for car park owners, claiming the pods could add $1 million to property values for "little or no investment." The company did not publish the basis for that figure.

Is Xeal first to put compute in car parks?

No. At least two other firms have moved into the same niche:

• Auddia, an AI music platform that pivoted to data centers, said in March it will deploy solar-powered GPUs in parking lots owned by a medical real estate firm, with a pilot set for the Dallas area.
• Belgian startup Tonomia announced last year a partnership with UK hardware provider Panchaea on eCloud, a distributed AI platform housed in solar canopy systems over parking lots.

Xeal differentiates by drawing on grid power already permitted at its own charging sites rather than adding generation.

The model still faces an open question: how much of the claimed 90 percent spare capacity holds up under sustained GPU load at scale. Xeal's 200MW figure covers capacity permitted and installed across all 1,600 locations, not capacity contracted to compute buyers.

For now, the number that matters is 100,000 GPUs.