Xeal CEO Nikhil Bharadwaj is looking at a different bottleneck while new data centers are still being built: power that is already connected but mostly unused.
Xeal says its U.S.
EV charging network has more than 1,600 locations and 200 MW of permitted electrical infrastructure.
Those charging sites typically operate at less than 10 percent of permitted capacity.
The company plans Latient Pods that can hold up to 48 Nvidia Hopper or Blackwell Ultra GPUs.
The point is to bring inference compute online at sites where the grid connection already exists.
For a small AI team, the scenario is familiar.
A product is ready for a real test, but capacity is far away or the next infrastructure decision takes longer than the experiment itself.
Compute located closer to users can shorten that wait and make lower latency testing possible.
Affordable AI compute gives teams another path while the big buildout catches up.
Run the test, measure what matters, and decide what deserves more capacity after the evidence is in.
BHK Cloud provides RTX 3090 compute at $0.15/hr.
Buy Now Pay Later means zero upfront, then payment after one month.
Which delay has held back your last AI test, compute capacity or getting it close to users?
https://ai.bhkcloud.com/?utm_source=linkedin&utm_medium=social&utm_campaign=bhk_social_2026w41&utm_content=evening
Originally posted on linkedin_personal