How are Data Centers Cooled? Liquid Cooling Technique
The short version: Siddhant Mehta and FluixAI's Abhi unpack why AI's GPU boom is forcing data centers toward liquid cooling — a shift with real power and water costs for host communities.
Our Take
Credit to Siddhant Mehta and his guest Abhi from FluixAI for pulling back the curtain on what's actually inside these buildings. Liquid cooling isn't just an engineering upgrade — it's an admission that the old chilled-air, water-hungry model can't keep up with AI-era chip density. That matters because every technical fix has a resource footprint, and right now the public rarely gets a say in which footprint their town absorbs.
The video is right to flag power redundancy and grid strain alongside cooling, because these systems don't exist in isolation — they're stacked on distribution infrastructure that, in a lot of counties, was never built for hyperscale loads. When a developer says 'we're switching to liquid cooling to save water,' the follow-up question communities should be asking is: save water compared to what baseline, and who's verifying it? Efficiency claims made in a webinar aren't the same as numbers reported to a local utility board.
We'd rather see this kind of technical literacy paired with local accountability — the permits, water allocations, and tax abatements that let a facility get built in the first place. Check our facility map to see what's landed near you, and if a project is moving through your county right now, our take-action page has steps for getting real answers before the concrete gets poured.
This is GridWatch the USA’s original commentary. The video above is the work of Siddhant Mehta, published on YouTube — full credit to the creator.