英伟达发布DSX POD模块化AI工厂设计
10月21日在GTC柏林大会上，英伟达发布DSX POD参考设计，由合作伙伴在模块化集装箱内预装、集成并验证完整的Vera Rubin AI工厂。英伟达称，在场地已具备电力、冷却、网络等配套基础设施的前提下，部署时间可从六个月缩短至最快14天。公司表示，这可能帮助客户更早开始生成生产环境Token并获得收入；合作伙伴计划于2027年初开始在全球交付。参与方包括Anduril、Armada、Bull、Crusoe、DayOne、戴尔科技、台达电、伊顿、鸿海、技嘉、HPE、联想、和硕、广达、施耐德电气、维谛和纬创。英伟达还发布了DSX OS，这是其内部数据中心运营软件的开源模块化开发者预览版。
该设计把集成和验证工作前移到制造工厂，从而减少部署现场的组装作业与协调工作。多篇报道将其视为英伟达把AI工厂本身打造为产品的举措，而部署速度、电力、安全和企业工作负载正成为基础设施建设的重点。
News Summary:
- NVIDIA introduces DSX POD, a reference design for complete Vera Rubin AI factories built by partners in modular containers, integrated and validated before delivery.
- Factory-built units reduce deployment time at a prepared site from six months to as little as 14 days, helping customers begin generating production tokens and earning revenue sooner.
- A global ecosystem of computing and infrastructure partners are working with NVIDIA to deliver modular AI factories worldwide in early 2027.
GTC Berlin—NVIDIA today introduced DSX POD, a reference design for complete NVIDIA Vera Rubin AI factories built by partners inside modular containers, integrated and validated before delivery. The units arrive ready to connect out of the box to a prepared site’s power, cooling, networking and other supporting infrastructure, reducing deployment time from six months to as little as 14 days.
DSX POD moves the integration and installation of computing, networking, cabling, rack-level power and liquid cooling to the manufacturing facility, reducing the extensive planning, construction, specialized labor and logistics of a traditional onsite build. Factory-tested containers reduce exposure to component delays and costly, error-prone onsite work, creating a shorter path to model training and production inference so customers can begin generating tokens and earning revenue sooner.
#NVIDIA#DSX POD#DSX OS#Vera Rubin NVL72#Dell Technologies#HPE#Lenovo#Schneider Electric#Vertiv#Bull#Foxconn#Quanta Cloud Technology