After using AI, companies seem to be poorer 用了 AI 之后，公司好像更穷了

Enterprises are caught in a collective self-hypnosis: the louder they chant "AI transformation," the more glaring the profit black holes in their financial reports become. The latest evidence comes from a trending headline, "After Using AI, Our Company Seemed to Get Poorer"—a title that’s like a bucket of cold water thrown on the frenzy.

Look at those earnings calls: CEOs scramble to report "how many AI tools have been deployed" and "how many large models have been built," as if these numbers alone could directly convert into stock price gains. But the reality is, GPU clusters are roaring, API call fees are skyrocketing, yet actual business process optimization is moving at a snail's pace. Many companies' AI strategies have essentially become expensive "tech decoration projects": wrapping the oldest processes with the latest algorithms, only shifting costs from Department A to Department B. Even more ironic is that those who advocate "AI replacing human labor" may now be paying double—shelling out for astronomical computing bills while still maintaining legacy systems.

On the same day, another trending piece, "Trillion-Dollar AI Companies Ban AI in Interviews," peeled back this glistening veneer. It’s practically the year’s best dark humor: on one hand, they sell the narrative that "AI will disrupt everything" to the world; on the other, in their own talent selection, they treat AI as a cheating tool. This schizophrenia reveals the industry’s deepest hypocrisy—they know exactly where the current boundaries of AI’s creativity lie, yet still choose to package it as an omnipotent myth. Why ban AI in interviews? Because those trillion-dollar companies understand that core innovation ultimately requires leaps of the human brain, not the stitching together of probabilistic models.

Even more intriguing is the capital market’s reaction. Just as the "AI money-burning" debate was raging, the Shenzhen Stock Exchange included AI chip and algorithm companies like Zhipu and Biren in the Hong Kong Stock Connect scheme. This move is like a magician’s sleight of hand: with the left hand, they reveal the truth of "industry losses," while the right hand immediately catches market sentiment with the story of "future tracks." But is capital truly blind? Of course not—they understand perfectly well. It’s just that in this game of musical chairs, as long as the music hasn’t stopped, no one wants to be the first to leave the dance floor.

Ultimately, we are experiencing a classic "tech hype cycle": after the period of inflated expectations, the trough of disillusionment inevitably arrives. Those companies that treat AI as a "plug-and-play solution" will hit the wall first, while those that survive will be the pragmatists who view AI as a "tool requiring fine-tuning" rather than a "philosopher’s stone." The future divide will emerge here: whether to continue paying a premium for concepts or start calculating the true ROI of every AI application.

The biggest irony is that while the entire industry is discussing how to use AI to "cut costs and boost efficiency," the first verified "efficient application" turns out to be—helping companies lose money faster. This feast doesn’t need more bubbles; it needs a sobering slap.

企业们正陷入一场集体自我催眠：把“AI转型”念叨得越响亮，财报里的利润黑洞就越刺眼。最新证据来自热榜那条“用了AI之后，公司好像更穷了”——这标题简直是给狂热泼的一盆冰水。

看看那些财报电话会议吧，CEO们争先恐后地汇报“已部署多少AI工具”、“建立多少大模型”，仿佛这些数字本身就能直接兑换成股价。但现实是，GPU集群在轰鸣，API调用费在飙升，而实际业务流程的优化却像蜗牛爬行。许多公司的AI战略，本质上成了昂贵的“技术装饰工程”：用最新的算法包装最旧的流程，结果只是把成本从A部门搬到了B部门。更讽刺的是，那些鼓吹“AI替代人力”的企业，现在可能正为天价的算力账单和依然需要维护的老旧系统支付双重费用。

而就在同一天，热榜另一条“万亿美元AI公司面试时禁止使用AI”则揭开了这层华丽的外衣。这简直是年度最佳黑色幽默：一边向全世界兜售“AI将颠覆一切”的叙事，一边在自己的人才选拔中把AI视为作弊工具。这种分裂暴露了行业最深层的虚伪——他们清楚知道当前AI的创造力边界在哪里，但依然选择将它包装成无所不能的神话。面试禁用AI？因为那些价值万亿美元的公司明白，核心创新终究需要人脑的跃迁，而不是概率模型的缝合。

更值得玩味的是资本市场的反应。就在“AI烧钱论”甚嚣尘上时，深交所把智谱、壁仞这些AI芯片和算法公司调入了港股通。这手操作像极了魔术师的障眼法：左手倒出“行业亏损”的真相，右手立刻用“未来赛道”的故事接住市场情绪。但资本真的看不懂吗？他们当然看得懂——只不过在击鼓传花的游戏里，只要音乐没停，没人愿意先离场。

说到底，我们正经历一场典型的“技术炒作周期”：预期膨胀期过后，幻灭低谷必然到来。那些把AI当作“即插即用解决方案”的公司会率先碰壁，而真正能存活的，必然是那些把AI当作“需要精细调教的工具”而非“点石成金法术”的务实派。未来的分野将出现在这里：是继续为概念支付溢价，还是开始计算每个AI应用的真实ROI。

最讽刺的是，当整个行业都在讨论如何用AI“降本增效”时，第一批被验证的“高效应用”居然是——帮公司更快地亏钱。这场盛宴需要的不是更多泡沫，而是一记清醒的耳光。