The Hidden Risks and Mechanics of the AI Investment Boom

After 40 years investing, the speaker warns that the riskiest moment in markets is when "everyone's celebrating," not during crashes. In the current AI boom, early insiders at OpenAI have sold $6.6 billion in stock, SpaceX is preparing to go public, and Anthropic and OpenAI are targeting valuations "near a trillion dollars each." This wave of public listings is reminiscent of historic patterns: Capital Economics notes that "surge[s] in companies issuing new shares" often signal an equity boom's end.

The boom's mechanics are fueled by belief: companies' valuations are marked up based on the latest trade, not the arrival of cash. This confidence-driven loop parallels housing markets or historical tech bubbles. Now, the cycle crosses into reality as IPOs demand "real investors with real money." The expected IPOs of SpaceX, OpenAI, and Anthropic could raise "$200 billion," but the infrastructure spending behind AI is even larger. Amazon, Meta, Google, Microsoft, and Oracle collectively commit over $700 billion in capital expenditure for 2026 alone—almost double last year's spend.

This enormous capital needs to come from somewhere, and in absence of "a secret reservoir of fresh money," investors must sell prior winners (Apple, Tesla, Microsoft, index funds) to fund new AI stocks, leading to the industry cannibalizing itself. The speaker details how this creates a "closed loop"—invest money in AI companies, chips from NVIDIA, and hardware providers, and watch valuations climb with each cycle. Companies call this a "virtuous circle," but the loop masks weaknesses if "outside demand stalls."

Further risk has been transferred through changes in index rules. Traditionally, new listings wait at least "months—sometimes a year" before index inclusion, allowing prices to settle. NASDAQ recently altered rules to allow "wildly speculative, barely tested stock" immediate entry into index funds. Bloomberg Intelligence estimates index funds may be forced to absorb "19% of SpaceX's available public shares" for S&P-style funds and "24%" for Russell and NASDAQ funds. S&P Dow Jones Indices maintained the normal waiting period, but most retail investors hold funds that may not.

The speaker raises questions about the validity of reported AI company profits. Michael Burry argues that depreciation schedules are optimistic: AI chips are depreciated over 5–6 years, but "a top chip might really only be cutting edge for two to three years," leading to inflated profits. Burry estimates Oracle may overstate profits by "27%" and Meta by "21%." No proven verdict exists—CNBC could not independently confirm—but even partial correctness would mean the foundation beneath AI stocks is less solid than believed.

History shows real technologies (railways, internet) yield destructive bubbles when the late buyers fund the peak. The AI story is not a fraud, but its reality is the "bait" that draws in late investors, especially index fund holders. The speaker's own investing response is to avoid both shorting AI stocks (risking loss against real technology) and chasing IPOs (risking overpaying at peak prices). Instead, he urges:

• know the contents and rules of your index funds
• make intentional choices about risk
• let new companies prove themselves with time
• keep costs low and stay cautious when "everyone around you is cheering"

The riskiest moments "never feel risky," so careful research and due diligence are essential as the AI boom morphs into a unique investment risk.