The problem with the life support system overtook the Nvidia GB200 NVL72 and NVL36 server systems before the start of mass production
A recent issue has affected Nvidia's latest GB200 NVL72 and NVL36 server systems, which are equipped with the advanced GB200 compute accelerators for artificial intelligence applications. Shortly before mass production began, a serious problem was discovered in the liquid cooling system.
GB200 NVL72 systems include 18 1U nodes, each containing a pair of GB200 accelerators consisting of two Nvidia B200 chips and one 72-core Arm Grace processor. The entire system consumes about 120 kW and is equipped with a liquid cooling system and a single DC power bus. According to preliminary data, the cost of the GB200 NVL72 system will be $3 million.
According to TweakTown, leaks were found in cooling systems associated with third-party components. The leaks were discovered before mass production began, giving Nvidia time to fix the problems. Despite the threat of delivery delays, the product is expected to be delivered on time.
The incident has raised concerns among major cloud service providers. Taiwanese manufacturers such as Shuanghong and Qihong are ramping up production of components to provide Nvidia with alternative options. We are actively working to resolve the issue, and server cabinets with the corrected cooling system will soon begin shipping to customers.
