Jensen Huang Talks Data Centers in Berlin: Construction Speed Becomes Part of the Compute Race
At Nvidia GTC Berlin, Jensen Huang explained how Nvidia is accelerating data center construction. The video is short and points to a practical bottleneck: scaling compute depends not only on chips but also on the construction site.
TIME posted a video on X. According to TIME's tweet, Nvidia CEO Jensen Huang spoke at GTC Berlin and explained how Nvidia will accelerate data center construction.
GTC is Nvidia's developer conference, and the Berlin edition is aimed at Europe. For those who haven't been following, a quick background note: AI data centers are not ordinary server rooms. They need power, cooling, networking, and servers racked, and they need a large number of racks brought online in a limited time. Chips arriving is only one part of it. If the construction cycle drags on, compute capacity can't come online.
TIME's tweet focuses on “speed up data center construction.” In other words, Nvidia wants to be involved not only in compute hardware but also in the process of data centers from design to deployment. The specific plans are in the video, and the text information is limited. What can be confirmed is: Nvidia treats construction speed as a problem that needs to be solved.
This angle isn't flashy, but it's very practical. Model training and inference both have to run on physical facilities. If power, cooling, or networking gets stuck, it turns into queues and delays. Ordinary users may not care how racks are installed, but they will feel whether services respond quickly and whether new features take a long time to launch.
The video is vertical and comes from TIME's X account. For AI practitioners, they can watch the video to see which specific stages Nvidia discusses. For ordinary readers, one takeaway is enough: the compute race isn't only happening on chips; it's also happening on construction sites and in server rooms.
发布时间: 2026-10-30 02:38