NVIDIA’s Jensen Huang Refutes Dot-Com Bubble Comparisons, Claims AI Frenzy Driven by Genuine Demand

NVIDIA’s Chief Executive Officer, Jensen Huang, has addressed concerns about the artificial intelligence (AI) sector potentially experiencing a downturn akin to the infamous dot-com bubble. Huang asserts that the current demand for computational power is robust and distinct from the factors that characterized the rapid expansion and eventual collapse of the Internet boom.

NVIDIA’s CEO Highlights Rising Demand for Computing Power in AI Development

The parallels drawn between AI advancements and the dot-com era largely stem from perceptions about stock prices rather than underlying technological realities. Many observers mention the rapid ascent of NVIDIA’s stock valuation, which has surged dramatically in a short period. Despite this growth, Huang contests the notion of a bubble. In an interview featured in the Financial Times’ segment “The Minds of Modern AI”, he emphasizes the substantive demand for computing power fueled by advancements in AI infrastructure, particularly pointing to the improved utility of “dark fibre”.

Question: Are you concerned that we might not fully grasp the situation and that we could see a reckoning similar to the dot-com bubble? If so, what misconceptions exist about AI demand compared to that era?

Jensen Huang: During the dot-com bubble, much of the deployed fibre remained unused, referred to as ‘dark fibre’.In stark contrast, nearly all GPU resources today are actively utilized.

For context, ‘dark fibre’ describes the scenario in the late 1990s when telecommunications companies expanded their internet infrastructure extensively, anticipating a surge in demand that ultimately did not materialize. This overbuilding resulted in little to no return on investment. This historical lesson underscores the difference between artificial and genuine demand in technological expansion.

Huang argues that while the general public often equates AI with prominent applications like ChatGPT or image generation tools, the true evolution of AI extends far beyond these representations. He claims that AI has reached a stage where it can “effectively think and ground itself through research”, a capability that has yet to see wide adoption. Additionally, both the volume of inquiries generated by businesses and the scale of computing needs are accelerating, indicating that current demand is legitimate rather than artificially inflated.

NVIDIA's CEO holding a Grace Blackwell die — NVIDIA’s CEO holding a Grace Blackwell die | Image Credit: NVIDIA

While we refrain from declaring whether Huang’s perspective on computing demand holds merit amidst comparisons to the dot-com era, it is evident that artificial intelligence possesses substantial growth potential, necessitating advanced computational capabilities. This demand could be fulfilled by major players like NVIDIA, AMD, and Intel. However, challenges such as energy consumption and the ability of cloud service providers like Microsoft to effectively integrate these advanced AI chips will play a crucial role in the technology’s future.