Nvidia Shareholders Major Reason to Cheer After Huang Keynote
Nvidia CEO Jensen Huang used the company's GTC Conference in Berlin to pitch data centers as "AI factories" built around a new server rack. Management says the…
Nvidia (NASDAQ: NVDA) shareholders have a major reason to pay attention to CEO Jensen Huang's keynote at the company's GTC Conference in Berlin, held Oct. 20 through Oct. 22. Huang spent much of the event describing data centers as "AI factories" and spotlighting a new Nvidia product he says will anchor that shift.
The product is a comprehensive server rack designed to deliver full AI factory performance. It combines 72 of Nvidia's next-generation Vera Rubin graphics processing units (GPUs) with 256 new custom language processing unit chips. Huang also spoke about the importance of AI technologies for European countries and companies.
The racks are not built only around Vera Rubin GPUs. They also include the company's Vera family central processing units (CPUs). By integrating its own CPUs and other hardware, Nvidia is widening the share of AI compute spending it can capture.
Sales opportunity per gigawatt keeps growing
During the company's recent quarterly call, management said its sales opportunity per gigawatt of compute has expanded from $18 billion with Hopper to $25 billion with Blackwell. With the Vera Rubin family, it now stands at $40 billion.
CPU integration is a big part of that increase. Nvidia also expects the growing importance of AI inference to drive new market share gains. GPUs remain the crucial hardware for training advanced AI models. But high-performance CPUs can often handle the inference work needed to run those models, and they can do it more efficiently in terms of hardware and energy costs.
At the company's previous GTC event in March, Huang described the rack as built for exactly that job. "This is the AI future. This is where AI wants to go. It's designed for inference, this one workload. And this workload is what drives AI factories," he said.
The cost of waiting
In Berlin, Huang put numbers on the stakes for operators. He said a 1-gigawatt AI factory would cost about $50 billion to set up. He estimates such a facility would generate roughly $40 billion in annual revenue.
Delay is expensive, he argued. Bringing an AI factory online just three months late would mean losing $10 billion in potential sales.
That argument speaks to a live concern in markets. Worries about a valuation bubble in AI have made many stocks in the category volatile. Huang is pointing to a different risk: players that do not spend enough on AI compute could be left behind on both technology breakthroughs and revenue.
For Nvidia shareholders, the pitch is that each new product generation lets the company sell more per gigawatt of capacity, and that customers feel pressure to build quickly.
Source: finance.yahoo.com
