Bloomberg Economics 的全球情境分析工具 SHOK 显示，人工智慧估值的崩溃可能会使全球 GDP 削减 1.3% 或 $1.6 trillion（约 1.6 兆美元）。这场危机将始于信心崩溃和美股暴跌 20%，导致信用利差扩大，且由于人工智慧基础建设停工，对资本支出造成每年 3% 的拖累。台湾、南韩和美国将是最大的输家，美国经济将被推向衰退边缘，经历两个季度的收缩，2027 年的 GDP 增长率将下降 1.5 个百分点，降至仅 0.4%。

这场冲击将在全球传播，严重影响依赖半导体的经济体。由于对 Taiwan Semiconductor Manufacturing Corporation、Samsung Electronics 和 SK Hynix 晶片的需求暴跌，台湾的年度 GDP 将受到约 4% 的打击，南韩则将超过 2%。相比之下，随著美国消费者紧缩支出和情绪恶化，中国和欧洲面临较小但仍显著的冲击。基于标准的总体经济关系，该模型表明美联储将降息三到四次，其他中央银行也将步入刺激模式。

最终，金融系统将面临违约增加和不确定性，这更像 2000 年的网路泡沫破裂，而非 2008 年的系统性危机。虽然人工智慧最终可能实现其生产力承诺，但历史经验表明，变革性技术通常需要数年甚至数十年的时间才能产生可衡量的生产力增长。如果人工智慧泡沫破裂，它确实会成为工作杀手，但这是通过投资情绪逆转引起的总体经济衰退，而非科技悲观主义者所担忧的技术性失业。本分析由 Jamie Rush、Rana Sajedi 和 Hyosung Kwon 撰写，并在 Michael Deng、Chris Collins、Bjorn Van Roye 和 Tom Orlik 的协助下完成。

Bloomberg Economics' global scenario analysis tool, SHOK, shows that a collapse in AI valuations could slash global GDP by 1.3% or $1.6 trillion. The crisis would begin with a collapse in confidence and a 20% plunge in US equities, leading to wider credit spreads and a 3% annual drag on capital spending due to a halt in AI infrastructure buildouts. Taiwan, South Korea, and the US would be the hardest hit, with the US economy pushed to the edge of recession, experiencing two quarters of contraction and a 1.5 percentage point drop in GDP growth for 2027 to just 0.4%.

The shock would propagate globally, heavily affecting semiconductor-reliant economies. Taiwan's annual GDP would be hit by about 4%, and South Korea's by over 2%, due to cratered demand for chips from Taiwan Semiconductor Manufacturing Corporation, Samsung Electronics, and SK Hynix. In contrast, China and Europe would face smaller but still significant impacts as US consumer demand shrinks and sentiment sours. Grounded in standard macroeconomic relationships, the model suggests the Federal Reserve would cut interest rates three or four times, accompanied by stimulus from other central banks.

Ultimately, the financial system would face defaults and increased uncertainty resembling the 2000 dot-com crash rather than the 2008 systemic crisis. While AI may eventually deliver productivity gains, history shows that transformative technologies require years or decades to manifest them. If the AI bubble bursts, it will indeed kill jobs, but through a macroeconomic downturn caused by a reversal of investment sentiment rather than the technological displacement feared by tech-pessimists. The analysis was prepared by Jamie Rush, Rana Sajedi, and Hyosung Kwon, with assistance from Michael Deng, Chris Collins, Bjorn Van Roye, and Tom Orlik.