This insight is pretty interesting. AI is massively disruptive to our era, just like how the internet was a huge disruption to the world back then. During the internet boom, a mainstream narrative was that data traffic would grow explosively, bandwidth would never be enough, and networks would need constant expansion, with base stations continuously being built. What actually happened was that the pace of the real world is always slower than the narrative. Demand did grow, but not as fast as predicted. And in the infrastructure industry, once utilization rates drop: the capital return model collapses, valuation systems reset, and stock prices plummet. Could the development of AI and computing power over the past few years follow a similar pattern? Computing power might get cheaper and cheaper, competition in large models could become too intense, and capital returns might collapse. But what I want to say is that we should still try to get involved as early as possible. Pessimists see the internet bubble bursting and stocks dropping 99% afterward. Optimists see Ericsson growing 1,000x from 1988 to 2000 in just 13 years, and NVIDIA growing 10x in nearly 5 years since the rise of AI. Thinking about when the bubble will burst is meaningless; instead, you might miss out on the entire era.