Jensen Huang stated at the Goldman Sachs Communacopia + Technology conference that NVIDIA is confident about continuing to expand its business next year, with revenue expected to grow by 70% year-over-year. This judgment continues the guidance provided by the company in its financial report last month and also reiterates the market's focus on the demand for AI infrastructure.
The core reason he gave is that NVIDIA is no longer just a single chip supplier; instead, it is deeply integrated into multiple links of the AI industry chain. Jensen Huang stated that from model companies and cloud service providers to original equipment manufacturers, and then to data center project construction, NVIDIA can observe relatively complete changes in demand.
The price of a single system is now much higher than that of traditional GPU.
Jensen Huang stated that the outside world is still accustomed to thinking of NVIDIA as a company that sells chips, but what the company delivers now are large-scale AI computing systems. He gave an example, saying that a system connected through NVLink contains about 2 million components, has a power consumption of up to 250,000 kilowatts, and is worth approximately 8.5 million US dollars per unit.
He also mentioned that a computing system composed of 36 Grace CPU and 72 Blackwell GPU is currently seeing monthly sales growth of 27%. This indicates that NVIDIA's current growth is not solely due to the shipment of individual chips, but rather more so from complete systems and cluster-level products.
The company claims to have covered the main demanders in the AI ecosystem.
Jensen Huang stated that NVIDIA's products are used by various model developers, including Anthropic, OpenAI, Google, as well as open-source model providers. He mentioned that NVIDIA has become one of the foundational platforms for the AI ecosystem and the AI industry.
He also stated that the company continuously monitors the progress of global data center projects, including the construction of land, power supply, and data center enclosures. According to him, feedback from emerging cloud providers, OEM manufacturers, cloud platforms, and AI native companies enables NVIDIA to grasp the trends in demand more promptly.
Responding to Questions about "Circular Trading"
Regarding the issue of whether there is a "circular transaction" between investment and procurement that has attracted external attention, Jensen Huang stated that before investing in related companies, NVIDIA would first confirm that the companies in question already have real customer contracts and are capable of generating revenue. He mentioned that he has seen related contracts totaling approximately $100 billion, and therefore believes that these collaborations have a clear commercial foundation.
However, whether NVIDIA will be able to maintain its current dominant position in the long term remains to be seen. As large technology companies such as Amazon, Microsoft, and Google advance with their own AI chips, and Anthropic, OpenAI, and other AI companies are also laying out their own hardware, the market is still assessing the impact of intensified competition on their growth.
