Nvidia's Q3 financial report is explosive: revenue of $57 billion accelerated quarter-on-quarter, with data center business soaring by 66%. CEO Jen-Hsun Huang opened by criticizing the AI bubble theory, claiming to see a "completely different picture".

But there is an interesting contrast here - many people are looking at Nvidia for AI bubble signals, but they are actually focusing on the wrong target. Nvidia is just laying the tracks; what we should really be looking at is the passengers.

OpenAI's revenue has surged to a run rate of $13 billion this year, with a valuation of $500 billion, and the growth rate is still doubling. Anthropic aims for a revenue run rate of $20-26 billion, with a valuation of $35 billion, which is not unreasonable. The growth data of these two AI software companies is the barometer for the risk of a bubble.

As long as OpenAI/Anthropic continues to grow at 2-3 times the speed, the positive cycle of AI adoption remains intact. Nvidia stock rebounded 6% that day, adding 300 billion in market value, but the issue is not with the chips; it is whether the backend AI applications are truly being used. That is the question that needs to be asked.