Thinking that AGI is unachievable and therefore AI is useless might be a misconception. It overlooks the fact that current AI capabilities are already having a massive impact across various fields. The demand is skyrocketing. Even without AGI, the token generation volume for reasoning has grown tenfold within a year, and this growth rate is still accelerating. From this perspective, as AI capabilities continue to penetrate production and daily life, the growth in computing demand could sustain for quite a while. Recently, there’s been a noticeable shift in AI—people have started using metrics like daily active users (DAU) and monthly active users (MAU) to evaluate major models or AI products. This essentially brings us back to the business model framework of the mobile internet era. One thing I’m personally very certain about is that the transformation driven by AI is accelerating—and the acceleration itself is accelerating. Yet, many people haven’t noticed or are completely unaware of it. A friend once said something that left a deep impression on me: 'Maybe AGI ultimately won’t be achieved, but we’ll all become dependent on large models.' I couldn’t agree more. Of course, I personally believe AGI can still be achieved—it’s just a matter of time.