Capital Economics: One AI Bubble Has Already Burst — The Next "Rare" One Is Still Growing

The First Bubble: Already Burst (估值泡沫)

According to Capital Economics chief markets economist John Higgins, the AI stock bubble debated throughout late 2025 has already deflated. He explained that "if you're judging whether a bubble exists or not in relation to how stretched or otherwise its valuation is, then there's an argument that the bubble has burst."

Evidence of the burst: - Tech sector price-to-earnings ratios peaked near 75% in late 2024 but fell by October 2025 to their smallest level since the pandemic - The pattern mirrors the dotcom bubble, though less severe (early 2000s IT sector exceeded 150% P/E ratio) - Software-as-a-service stocks suffered significantly during the "SaaSpocalypse," with Salesforce and ServiceNow each losing approximately 30% of value since year-start - Semiconductor industry experienced recent slowdowns due to chip shortages and geopolitical tensions

The Second Bubble: A "Rare" Kind Still Growing (盈利泡沫)

Higgins identified an unusual bubble forming in fundamentals rather than valuations. He stated: "there may be one [bubble] actually in the fundamental side of things, which is quite rare...the bubble actually may be in the earnings themselves."

Key concern: Whether extraordinarily high tech earnings remain sustainable.

Supporting data: - Magnificent Seven companies show estimated earnings growth around 18%, versus 11% for remaining S&P 500 companies - Nvidia reported $68.1 billion fourth-quarter revenue — a 73% year-over-year increase - 498 AI unicorns valued at $2.7 trillion exist as of fall 2025 - OpenAI's valuation jumped from $500 billion in October to $730 billion in March

Risk factors for earnings cliff: - AI demand may be lower than anticipated despite $539 billion in projected 2026 capex - Employee anxiety about displacement is stalling adoption despite 88% of companies reporting regular AI use - Economic weakness could suppress demand, particularly given helium supply disruptions from Qatar affecting chip manufacturing and data center vulnerabilities during regional conflicts