2025年的爆款成功显示出三项可量化风险：不可预测的扩张、影子市场，以及快速跟进的竞争者。OpenAI称约有全球十分之一的人口使用ChatGPT，并预测到2030年年收入约为2000亿美元（约为当前年化水平的10倍），且已承诺投入1.4万亿美元用于算力，因此一旦增长路径不及预期就可能资不抵债并使AI热潮戛然而止。

投资不足会催生仿制品：尽管诺和诺德的Wegovy短缺已于2月结束，约有100万美国人仍在使用更便宜的复方仿制品，而诺和诺德称其更不安全；同时泡泡玛特在与拥有八颗或十颗牙齿的假Labubu作斗争。先发劣势也体现在肥胖药物上：礼来公司的Zepbound比诺和诺德晚两年进入美国市场却在2025年领先；在AI领域，继DeepSeek在1月以极低成本发布后，谷歌推出与OpenAI最佳模型势均力敌的模型，并凭借覆盖近40亿安卓用户的分发能力加剧竞争压力。

更安静的规模优势看起来更持久：沃尔玛的市值在成本杠杆与数字化改造吸引手头紧张的消费者之际升至接近1万亿美元。宁德时代通过高强度研发巩固了领先地位，并利用其5月在香港的二次上市——2025年全球最大的股票发行——为向电网电池扩张融资，从而凸显可复制的爆款很少能转化为持久优势。

OpenAI states that about one-tenth of the world uses ChatGPT, and they project an annual revenue of roughly $200 billion by 2030, which would be about ten times today's rates. They have also committed $1.4 trillion to computing, so failing to meet growth expectations could lead to insolvency and stall the AI boom.

Now, underinvestment leads to competition. For example, even though Novo Nordisk's Wegovy shortage ended in February, around 1 million Americans still resort to cheaper compounded versions that Novo considers less safe. Additionally, first-mover disadvantages appear in the obesity drug market, where Eli Lilly’s Zepbound, coming two years late, is challenged by strong competition.

Meanwhile, CATL is solidifying its leadership through significant R&D efforts, and its upcoming May secondary listing in Hong Kong is expected to be the largest share offering in 2025. This funding will assist in their expansion into grid batteries, emphasizing that replicable successes often don’t lead to lasting advantages.

Source: What Novo Nordisk, OpenAI and Pop Mart have in common

Subtitle: All three have suffered the curse of overnight success

Dateline: 12月 18, 2025 07:37 上午