Today, let's talk about two most noteworthy trends: AI commercialization has entered deep water, and the Renaissance movement of the old Internet.
Let's talk about the commercialization of AI first. GitHub Copilot was scolded for changing its billing model, Sige New Energy developed the so-called global AI agent, and SoftBank spent 75 billion euros to build AI facilities in France. These few things are very interesting when viewed together. The Copilot case directly exposes the dilemma of the commercialization of AI products-using free strategy to circle users in the early stage, and suddenly harvesting when users rely on them. This trick has long been rotten in the SaaS field. However, the particularity of AI products is that the consumption of tokens is completely uncontrollable, and developers may burn hundreds of dollars while debugging. This is not a user education issue at all, it is the original sin of product design. In contrast, Sige New Energy's AI agent is even more ridiculous. It is directly bound to the just-needed scenario of energy management, at least it can be calculated. Softbank's crazy investment is at the other extreme. Smashing AI infrastructure now is like smashing optical fiber 20 years ago. The bet is that AI computing power will become infrastructure like hydropower in the next decade. But the problem is that most AI applications today simply cannot support such high infrastructure utilization, and it is likely to turn into another capital bubble.
The resurrection of Tianya Community is particularly magical. In an era when algorithm recommendations and Short Video rule, engaging in retro BBS is like selling BPs in the era of smartphones. But think about it carefully, young people are now beginning to miss the "slow social networking" of the early Internet. This is also the logic that forums such as Reddit have recently skyrocketed valuations. Tianya's real opportunity is not to replicate its past glory, but to become a neutral platform for anti-algorithm and anti-fragmented content. Just like the renaissance of vinyl records in the digital music era, what sells is not sound quality but emotion. But the risks are also obvious-that group of users had long been domesticated by WeChat Weibo. Why should new users give up the instant enjoyment to accompany you in slow social networking? Unless you can find vertical communities such as game players like Discord, there is a high probability that you will become a niche.
Finally, let's make an observation: AI tracks have now been clearly divided into infrastructure and application schools. The infrastructure faction is in the center of crazy construction computing power, and the application faction is looking for various strange scenarios to implement them (even pet robots are not spared). This is very similar to the situation during the Internet bubble in 2000-when telecom companies were struggling to lay optical fibers and startups were making up stories. History will not simply repeat itself, but rhyming is for sure. Now, people entering AI applications should first see if they can withstand the critical hit of the cloud service provider's bills.