AI ( Artificial Intelligence ) Nvidia, the chip leader The CFO refuted the "AI bubble" claim, emphasizing the company's leading edge.
On December 2nd local time, Nvidia CFO Colette Kress attended the UBS Global Technology and Artificial Intelligence Conference and gave an interview. Regarding the issue of an "AI bubble," Kress stated that most of Nvidia's new AI chips currently being delivered are... All of them are used to add new data centers The shift towards new infrastructure, rather than replacing existing capacity, means that new computing power is constantly being added, and the global economy is in the "early stages" of transitioning to the data center infrastructure needed for AI.
Kress reiterated the company's previous forecast: Nvidia expects global AI investment to reach $3 trillion to $4 trillion by the end of 2030, with about half of that amount related to the shift from traditional CPUs to accelerated computing.
When asked if Nvidia's lead was narrowing, Kress responded, "Absolutely not": "It's important to emphasize that we have a highly collaborative co-design of all aspects of our GPUs... This has nothing to do with, and is completely different from, any fixed-function ASIC on the market. Currently, everyone is using our platform. All models run on our platform, whether deployed in the cloud or on-premises, and all workloads do too."
Previously, reports surfaced that Meta was considering deploying Google's self-developed ASIC TPUs (Tensor Processing Units) in its data centers starting in 2027, a deal worth billions of dollars, causing Nvidia's stock price to fall by more than 7% intraday. In response, Nvidia publicly stated that its GPUs "remain a generation ahead of the industry."
Regarding customer profitability, Kress stated that as the AI industry transitions from generative models to inference models, the scale of models increases, the amount of tokens generated increases, and users will be more willing to pay, creating a "flywheel effect." She pointed out that the computing power needs of many large model developers are more of a long-term issue that needs to be addressed gradually based on capital availability.
It's worth noting that in September of this year, Nvidia announced it would gradually invest up to $100 billion in OpenAI. According to Kress, the project is still in the letter of intent stage and has not yet been finalized, but she did not specify whether there were any obstacles: "We haven't reached a final agreement yet, but we are working with them."
Furthermore, Nvidia CEO Jensen Huang stated that the company's total sales of Blackwell and Rubin chips would reach $500 billion by the end of 2026. Kress pointed out that Nvidia's agreement with OpenAI may result in the supply of chips to the latter, which is not yet included in the $500 billion sales forecast.
Kress also mentioned that Nvidia's commitment last month to invest up to $10 billion in OpenAI competitor Anthropic could further boost computing chip bookings.
On December 2nd, according to Dazhihui VIP. Nvidia (Nasdaq: NVDA) shares rose 0.86% to close at $181.46 per share, with a total market capitalization of $4.41 trillion. Nvidia's stock price has risen more than 30% year-to-date.
(Source: The Paper)
