Many observers believe the current artificial-intelligence boom is a kind of “bubble” that must eventually burst because of what they view as AI’s speculative nature and its detachment from real underlying value.
In an interview with The Telegraph, Nvidia executives strongly rejected suggestions that today’s AI boom resembles an inflated bubble. The discussion included comparisons to Enron – the energy giant that collapsed in the early 2000s – which Nvidia called completely inaccurate. The company argues that its value is not based on aggressive accounting or speculative assets but on the large-scale, real-world deployment of AI across science, industry, medicine and digital services. According to Nvidia, the market is still in an early stage of development, and demand for Hopper and Blackwell-series chips is growing faster than supply.
The company points out that the rapid expansion of AI data centers – along with progress in robotics, scientific computing and autonomous vehicles – means demand for compute power will only continue to grow.
But despite these assurances, Nvidia’s stock still experienced a sharp drop. CNBC reported that shares fell by as much as 6.7% at the steepest point of the session and later stabilized at -4.3%, after news that Meta is considering buying AI chips from Google to diversify its hardware suppliers for model training. The report raised investor concerns that even Nvidia’s key customers may begin to seek alternatives, especially given the growing competition from Google’s tensor processing units (TPUs), which could be attractive from a cost perspective.
Nvidia maintains, however, that demand for its AI accelerators is so strong that customers already have to queue for the hardware. The company argues that these market shifts – including major players developing their own chips and rising competition – do not signal weakened positioning, but rather the natural maturation of the AI-accelerator ecosystem. Leadership emphasizes that Nvidia’s technological edge remains significant, and the newest AI systems “consume” ever-increasing amounts of compute power, giving the company a solid business foundation.
Perhaps Nvidia is right, and we are only at the beginning of the AI development cycle – and the number of companies producing competitive AI accelerators will indeed grow. Still, investors reacted quickly, and even a giant like Nvidia cannot ignore signs of a potential shift in market dynamics. If other Nvidia customers begin investing in alternative AI accelerators, competitors will gain momentum, which will undoubtedly accelerate their development. And if Google’s TPUs prove highly effective in practice, Nvidia may have to speed up the evolution of its own AI-accelerator lineup even further.
