这个问题不是我凭空想到的。

This question did not come out of nowhere.

我以前在 Meta 和 Microsoft 工作过，所以对大公司有一种很复杂的感受：它们真的很强。不仅有 sharp 的头脑，还有强大的资源。强在人才、资源、分发、组织能力，也强在一旦方向确定，就能把某个东西快速推到全球用户面前。

I used to work at Meta and Microsoft, so my feelings about big companies are complicated. They are genuinely strong. They have sharp minds and enormous resources: talent, distribution, organizational capacity, and the ability to push something to a global user base once the direction is clear.

也正因为我见过这种强大，我反而更好奇：为什么今天这些公司面对 AI，会表现得这么紧张？

Because I have seen that strength up close, I became even more curious: why do these companies look so tense when they face AI today?

它们并不缺工程师，不缺数据，不缺钱，也不缺用户。按理说，如果 AI 只是一个更聪明的功能，比如帮你写一段话、生成一张图、总结一个文档，那大厂应该是最容易赢的。它们可以把 AI 塞进搜索、Office、广告系统、社交产品、云服务里，然后继续扩大原来的优势。现在大多数大厂尝试的也是这件事：把 AI 塞进现有产品里，加速，赋能，提升效率。

They are not short on engineers, data, capital, or users. If AI were only a smarter feature, something that helps you write a paragraph, generate an image, or summarize a document, big companies should be in the best position to win. They can put AI into search, Office, advertising systems, social products, and cloud services, then extend their existing advantages. In fact, most big tech companies are trying some version of this: putting AI inside existing products to accelerate, empower, and improve efficiency.

但 Meta 的大动作，让我感受到了另一层东西：大厂的恐惧。

But Meta's aggressive AI moves made me sense something deeper: the fear of incumbents.

它们恐惧的可能不是“某个功能被 AI 替代”，而是用户入口被重新改写。

What they may fear is not that one feature gets replaced by AI. It is that the user entry point gets rewritten.

拿 Google 举例。Google 最可怕的地方，不只是搜索技术，而是它长期占据了用户表达意图的入口。过去你有一个问题，第一反应是打开 Google，输入关键词，浏览网页，自己判断、拼接、比较，最后形成答案。Google 就是很多事情的第一步。

Take Google as an example. Google's most powerful position is not only its search technology. It has long owned the place where users express intent. In the past, when you had a question, your first move was to open Google, type keywords, browse pages, compare information, and assemble your own answer. Google was the first step for many things.

但 GPT 改变了这个动作。现在你可能不会先搜索，而是直接问：“帮我理解这个问题”“给我比较几个选择”“如果我是创始人，我该怎么判断？” 这时，AI 不只是给你答案，它接走的是中间那段认知劳动：理解问题、分析意图、组织信息、过滤噪音、提出下一步。

GPT changes that motion. Now you may not search first. You may simply ask: 'Help me understand this problem,' 'Compare these options for me,' or 'If I were a founder, how should I think about this?' At that moment, AI is not just giving you an answer. It is taking over the cognitive work in the middle: understanding the problem, analyzing intent, organizing information, filtering noise, and suggesting the next step.

换句话说，搜索开始变成基础设施，而 GPT 这种可以满足复杂需求的产品，开始成为入口。

In other words, search begins to look more like infrastructure, while a product like GPT, which can satisfy complex user needs, starts becoming the entry point.

这对 Google 来说，真正危险的不是少了几个蓝色链接，而是用户不再从“搜索”开始。

For Google, the real danger is not losing a few blue links. It is users no longer beginning with search.

OpenAI 的 GPT 让我看到的也是这一点。它强的地方不只是模型能力，而是它站在了用户意图的最前面。很多时候，人并不知道自己要什么，只知道自己有一个模糊的问题、一个没想清楚的判断、一个需要被整理的局面。之后，人们需要理解问题、分析意图、拆解路径，而 GPT 正好承接了这种模糊性。

OpenAI's GPT shows me the same pattern. Its strength is not only model capability. It stands at the front of user intent. Often, people do not know exactly what they want. They only have a fuzzy question, an unresolved judgment, or a situation that needs to be organized. Then they need to understand the problem, analyze intent, and break down possible paths. GPT is unusually good at receiving that ambiguity.

这就很关键了。

That is the key.

未来很多产品不会消失。Google 还会有数据，Microsoft 还会有 Office，Meta 还会有社交图谱。但它们可能从“用户第一站”，变成 AI 背后的执行层和资源层，而不是入口。

Many products will not disappear. Google will still have data. Microsoft will still have Office. Meta will still have social graphs. But they may move from being the user's first stop to becoming the execution layer and resource layer behind AI, rather than the entry point itself.

这正是让大厂恐惧的地方。不是 AI 会不会写文章、写代码、做 PPT，而是下一次用户产生一个问题时，第一个想打开的东西，不再是它们。

That is what truly scares big companies. It is not whether AI can write articles, code, or slides. It is whether the next time a user has a question, the first thing they want to open is no longer them.

这也是我最近一直在想的一个问题：在 AI 时代，真正的竞争可能不再是“谁拥有最多功能”，而是“谁拥有用户表达意图的那一刻”。因为谁拥有这个入口，谁就有机会重新组织整个市场。

This is the question I keep coming back to: in the AI era, the real competition may no longer be about who has the most features. It may be about who owns the moment when users express intent. Whoever owns that entry point gets the chance to reorganize the entire market.

之后，我又和 GPT 讨论了一下 AI-native 公司的竞争优势。如果大厂害怕入口被拿走，那么 AI-native 公司真正的机会，也许不是做一个更小的工具，而是从第一天开始，就围绕用户意图、认知劳动和最终结果来设计产品。

After that, I discussed the competitive advantage of AI-native companies with GPT. If incumbents fear losing the entry point, then the real opportunity for AI-native companies may not be building a smaller tool. It may be designing around user intent, cognitive work, and final outcomes from day one.