News
Nvidia CEO Jensen Huang says company may need to produce 200 server pods per week
Saturday, March 28, 2026 at 12:29 PM
Nvidia CEO Jensen Huang indicated that the company may need to produce approximately 200 GB200 NVL72 server pods per week to meet demand. Financial analysts at Wolfe Research noted that this production rate would significantly exceed current 2027 revenue estimates if taken literally, implying a monthly revenue of $120 billion.
Context
In a March 2026 appearance on the Lex Fridman Podcast, Nvidia CEO Jensen Huang stated that the company may need to produce approximately 200 server pods per week to meet the infrastructure demands of the agentic AI era. This comment followed the company's GTC 2026 announcement of the Vera Rubin architecture, which is optimized for autonomous AI agents. Wolfe Research analysts noted that while the figure may be contextual rather than a literal production target, the scale is significant; they calculated that 200 pods per week could imply a staggering $120 billion in monthly revenue.
This production scale represents a major leap for Nvidia's rack-scale computing strategy, which now includes the GB200 NVL72 and upcoming Rubin Ultra systems. With Wolfe Research estimating $150 million in Nvidia content per pod, these systems are becoming the primary growth engine for the company's data center segment. As Nvidia targets a $4.16 trillion valuation, the push for weekly triple-digit pod output highlights the massive capital shift toward liquid-cooled exascale clusters required for next-generation AI reasoning and agentic workflows.
Sources (9)
Top Technology PodcastsGB200 NVL72 | NVIDIA[PDF] Supermicro NVIDIA GB200 NVL72 SuperClusterHardware — NVIDIA DGX GB Rack Scale Systems User GuideWolfe Research reiterates Nvidia stock rating on agentic AI opportunity By Investing.comtranscripts Archives - Lex FridmanWhat is NVIDIA? History of Jensen Huang, CUDA & GPUs ...Lex Fridman Podcast - Podcast Analytics & Insights - Podscan.fm
Related Companies
Nvidia
NVDA