Nvidia Partner Vertiv Soars as Orders Double: “Evolving Faster Than Ever”

U.S. stocks fell, with the Nasdaq down 2.9%, as Intel and Amazon's earnings raised concerns about AI spending. Intel's stock dropped 28.6%, while Nvidia and AMD also declined. The S&P 500 and Nasdaq saw monthly losses, though the Dow ended October positively. Earnings season shows 82% of S&P 500 companies beating forecasts, with 31.6% expected YoY growth.

Nvidia reduced memory and interconnect bandwidth targets for its Vera Rubin AI hardware, with HBM4 bandwidth dropping 13% to 19.2 TB/s and NVLink 6 bandwidth falling 16.7% to 3 TB/s. The changes affect all configurations, including the 72-GPU rack. Nvidia stated the original figures were preliminary targets, and the new specs reflect production capabilities.

The US Justice Department is investigating Nvidia's $12.93B acquisition of Hugging Face, announced in September 2026, over concerns about market dominance in AI model distribution. The deal, expected to close in early 2027, involves a platform hosting 3M+ models and 500K datasets, with 18M developers and 200K+ customers. Nvidia faces scrutiny over potential competitive advantages and tied-selling allegations, while it pledges to keep the platform open to rivals.

Nvidia is investing $1 billion in Naver, gaining a 4.54% stake. The deal supports a $10 billion AI infrastructure project, with Naver's GAK Sejong data center expanding from 55 MW to 200 MW by 2028. Naver's entertainment assets may serve as a testbed for AI applications, offering quicker ROI than other industries. The investment is scheduled for November 6, with new shares listed on November 27.

Nvidia has reduced the memory and interconnect bandwidth specifications for its Vera Rubin GPUs, cutting per-GPU HBM4 bandwidth by 13% and NVLink bandwidth by 17%. The changes were made without a press release or explanation. The downgrades affect various configurations, with the eight-GPU HGX Rubin NVL8 experiencing a 26% bandwidth reduction. AMD's MI455X and Helios now lead in certain bandwidth metrics, though no benchmarked comparisons exist.

Google Cloud is deploying instances with Nvidia's Rubin GPUs, with one site supporting up to 80,000 units. This indicates Google's significant investment in Nvidia's architecture alongside its own Tensor Processing Units. Analyst Beth Kindig highlighted the scale as a key differentiator. (NVDA)