800V DC stayed in the lab [/ˈvi ˌdiˈsi ˈsteɪd ɪn ðə ˈlæb/] n - The idea. For a new AI cluster we examined 800-volt DC distribution from the substation all the way to the rack. The reckoning was attractive: higher voltage means lower current for the same power, which means smaller buswork and fewer conversion losses. A few hyperscalers were speaking of it publicly, so the team built a small lab bench to prove the concept.
What happened. [/wɒt ˈhæpənd/] n - The efficiency gain was real, but smaller than the model foretold once we counted the rectifiers, protection, and monitoring the true architecture would want. Worse, the equipment ecosystem was thin. Rack PDUs, breakers, busway sections, and safety gear fit for that voltage at our current densities were custom, unavailable, or sold only with long lead times and a single-source vendor.
Why we stopped. [/waɪ wiː ˈstɒpt/] n - The site could not afford to be the reference customer for that supply chain. A failed breaker or a delayed PDU would have silenced the whole row, and the people qualified to work on 800 V DC safely were not the people we already had on shift. The risk-adjusted cost stood higher than the savings.
What we kept. [/wɒt wiː ˈkɛpt/] n - The trial obliged us to model losses more carefully, and that labor was repaid in a conventional 400/230 V design with better rectifier placement. The concept remains on the roadmap for a future build, when the equipment market is riper, as recorded in The New Buildout.