NVIDIA CEO Huang Refutes AI Overinvestment Claims, Cites Growing Demand

NVIDIA CEO Jensen Huang directly challenged the notion that AI investment is excessive, arguing that the development of more practical and intelligent AI models will inevitably lead to higher demand for computing power. Speaking at the Wistron plant opening in Dallas, Texas, Huang stated that the increasing utility of AI models like China's Kimi K3 will attract more users, thereby boosting the need for the underlying computational resources.

“

The more intelligent and practical the AI model, the more people it will attract, thereby increasing computing power demand.

Huang explained that the efficiency gains from models like Kimi K3 do not signal a decrease in AI hardware demand. Instead, he sees these advancements as driving faster adoption by businesses and consumers. He views the emergence of open-source models as beneficial to the overall AI industry, asserting that both open and closed models, such as those from Anthropic and OpenAI, are necessary for a thriving AI ecosystem.

Looking ahead, Huang anticipates new business models emerging around AI computing power. He pointed to NVIDIA's computing platform as a stable, long-lasting resource that is easily rentable. As more companies adopt NVIDIA's platform, Huang suggested that idle computing capacity could be rented out, enhancing equipment utilization.

“

The market previously misunderstood DeepSeek, and now it is similarly misunderstanding Kimi K3.

Regarding the rapid expansion of AI infrastructure, Huang noted the industry-wide adoption of liquid cooling in next-generation AI data centers. This technology significantly reduces water consumption and improves energy efficiency compared to older evaporative cooling methods. He encouraged the market to replace older facilities with these new, more sustainable AI infrastructures.

“

The market will need both closed models from companies like Anthropic and OpenAI, and open models, to jointly promote the continuous expansion of the AI ecosystem.

Huang also acknowledged the substantial power requirements of AI data centers but framed it as an opportunity for the U.S. to upgrade its electrical grids, develop renewable energy sources, and build new power generation facilities. He believes AI capital expenditure can drive the modernization of urban infrastructure, benefiting power grids, energy systems, and public utilities. However, he also recognized community concerns about the rapid construction of AI facilities and recommended early collaboration between AI companies, local governments, and residents. He suggested that AI investments should also include contributions to public amenities like schools and parks, ensuring AI development benefits local communities.

“

Next-generation AI data centers have fully adopted liquid cooling technology.