NVIDIA & Foxconn To Build Taiwan's Fastest Supercomputer, Powered By Blackwell GB200 NVL72 Servers & Up To 90 ExaFlops 1

NVIDIA Ships First Blackwell GB200 NVL72 Server Racks Despite Thermal Concerns

NVIDIA’s much-anticipated Blackwell architecture is encountering some bumps on the road as thermal issues are reportedly plaguing its NVL72 server racks. As these server racks make their way to the mainstream market, a report from The Information highlights fresh challenges following the earlier hurdles related to interconnect technology.

Originally, problems with the interconnect technology led supply chain entities, notably TSMC, to reevaluate their manufacturing strategies, consequently affecting Blackwell’s availability. Now, the spotlight is on the NVL72 configuration, a vital component for NVIDIA due to its potential for revenue generation and client interest. Continuation of this issue might cast a shadow on NVIDIA’s server revenue, given the pronounced demand for the NVL72 model.

The current reports suggest complications arising from the cooling methods being implemented. In response, NVIDIA has prompted its suppliers to urgently address the situation as it significantly impacts the supply chain. However, with NVIDIA’s strong supply network and resources, a swift resolution seems probable as design revisions are already likely underway.

In an interesting development, Dell, a key partner with NVIDIA, has begun shipping PowerEdge XE9712 servers, which are the first enterprise-ready NVIDIA GB200 NVL72 servers. This advancement underscores the potential workaround and the gradual easing of supply constraints unless the design issues pose ongoing impediments.

Despite the present obstacles, the Blackwell lineup is still anticipated to be one of NVIDIA’s most successful launches, projected to generate significant revenue. With a favorable market shift towards AI clusters, demand for NVIDIA’s innovative architecture continues to rise, promising a bright future for the tech giant.

NVIDIA remains collaboratively engaged with top-tier cloud service providers as part of a strategic effort to refine these server solutions, anticipating successful iterations that will meet market expectations and advance AI infrastructure significantly.