泡沫不只在 AI。黄金在截至 12 月 12 日的一年内上涨近 64%；有人也担心政府债务与私募信贷（约 3 兆美元市场、部分资金流向 AI 资料中心）正在累积风险，甚至被形容为不透明、缺乏监管的「垃圾放贷」。加密资产的波动更极端：比特币市值年初至 10 月 6 日增加 6,360 亿美元，之后到 12 月 12 日不但回吐且跌得更多；迷因币成交额从 1 月 1,700 亿美元跌到 9 月 190 亿美元，$TRUMP 与 $MELANIA 自 1 月 19 日以来分别下跌 88% 与 99%。
作者把这种投机描述为「赌场化」的注意力经济：资讯同步扩散带来模仿与群聚。数据也显示致富焦虑升高：Harris 民调称 6 成美国人向往「极端财富」；Z 世代与千禧世代 70% 想成为亿万富翁，对比 X 世代与婴儿潮的 51%；Empower 研究显示，Zoomers 认为「财务成功」需要接近 60 万美元年薪与 1,000 万美元净资产。在 AI 基建上，大型科技公司尚可用资产负债表承受，但较弱的参与者更依赖杠杆，例如甲骨文为在德州与威斯康辛建资料中心而举债 380 亿美元，若缺乏真实客户，泡沫外溢风险可能扩大。
The essay uses the 1929 “Babson Break” as a parallel to today’s bubble psychology: warnings are easy to dismiss when optimism and FOMO dominate. In AI, Omdia estimates tech firms are on track to spend just under $1.6 trillion per year on data centers by 2030, even though AI’s profit case is still largely hypothetical and valuation expectations are racing ahead of fundamentals.
Bubbles appear across assets. Gold rose almost 64% in the year to Dec. 12; concerns also circle government debt and private credit, a roughly $3 trillion lending market that helps fund projects like AI data centers and is criticized as opaque and lightly regulated. Crypto shows the most quantifiable whiplash: Bitcoin’s market value gained $636 billion from the start of the year to Oct. 6, then gave back all of that and more by Dec. 12; memecoin volume fell from $170 billion in January to $19 billion by September, while $TRUMP and $MELANIA have dropped 88% and 99% since Jan. 19.
The piece links these swings to an attention-driven, casino-like investing culture amplified by TikTok, chats, and Reddit. Polling and studies suggest rising wealth obsession: a Harris poll finds 6 in 10 Americans aspire to “extreme wealth”; 70% of Gen Z and millennials want to be billionaires versus 51% of Gen X and boomers; Empower reports Zoomers equate “financial success” with nearly $600,000 salary and $10 million net worth. In AI infrastructure, big tech may absorb setbacks, but leveraged players heighten systemic risk—Oracle, for instance, is raising $38 billion in debt for data centers—reviving fears of debt-fueled overbuilding without real customers.
