NVIDIA CEO Jensen Huang Says AI Demand Is Real, Not a “Bubble,” as Compute Power Rises Alongside Query Volume
NVIDIA CEO Jensen Huang believes that the ongoing surge in artificial intelligence infrastructure demand is not a speculative “bubble,” but a reflection of genuine, sustained growth in computing needs. In an in-depth discussion during the Financial Times segment “The Minds of Modern AI”, Huang compared today’s AI revolution to the dot-com era, arguing that while both share similarities in market enthusiasm, the underlying demand today is fundamentally real.
When asked whether the AI boom could face the same fate as the dot-com bubble, Huang drew an analogy to the concept of dark fibre. During the late 1990s, telecom companies heavily overinvested in laying optical cables, anticipating exponential internet traffic that never materialized, leaving vast amounts of unused “dark” fiber optic infrastructure. “During the dot-com era, during the bubble, the vast majority of the fiber deployed were dark, meaning the industry deployed a lot more fiber than it needed,” Huang explained. “Today, almost every GPU you could find is lit up and used.”
The CEO emphasized that, unlike the speculative overbuild of the 1990s, the current explosion in AI infrastructure is driven by real, measurable workloads. Every available GPU deployed in data centers is running large-scale AI models and serving increasingly complex queries. Huang noted that computing demand is rising in direct proportion to the number and sophistication of AI queries being generated, highlighting that AI systems today do not just respond but are capable of reasoning, analyzing, and “grounding themselves through research.”
In Huang’s view, mainstream consumers often see only the surface applications of AI, such as chatbots and image generators, without realizing the depth of computation powering them. Behind the scenes, data centers are expanding capacity at an unprecedented rate to meet the scaling needs of advanced AI models, further validating the real demand behind the sector’s rapid growth.
While NVIDIA remains a central figure in the AI hardware ecosystem, Huang’s remarks underline a larger industry reality: the demand for compute power extends beyond a single company. Competitors like AMD and Intel are racing to develop next-generation AI accelerators, and hyperscalers such as Microsoft, Google, and Amazon are working to integrate these chips into their own cloud ecosystems.
However, this surge comes with challenges. The energy consumption required to sustain massive AI compute clusters has become a pressing global issue. The question is no longer whether AI demand is real, but whether infrastructure and energy systems can scale responsibly to meet it.
Huang’s comments ultimately reinforce NVIDIA’s position that AI represents not a fleeting market bubble but a structural shift in how industries compute, innovate, and interact with data.
Do you agree with Jensen Huang that AI demand is genuine and not a bubble, or are we heading toward another tech overbuild like the dot-com era? Share your perspective below.
