AI Bubble Math: 15% of U.S. GDP by 2030 to Break Even?

Here is a number that should make you put down your coffee: 15%. That is the share of U.S. GDP that Chris Clothier of CG Asset Management says AI users would need to spend on AI services by 2030 just to justify what hyperscalers are spending on infrastructure right now. He laid this out on the Bloomberg Merryn Talks Money podcast, and he called it, in his words, completely implausible. Hard to argue with that. For context, the entire U.S. healthcare system runs at roughly 17% of GDP. So we are talking about AI services becoming almost as economically dominant as keeping people alive. Sure.

And it is not just one contrarian fund manager sounding the alarm. Capital Group's Jody Johnson, flagged in Bloomberg Daybreak's Asia coverage, pointed to a real and growing gap between AI company expectations and actual enterprise adoption rates. She also highlighted something that does not get enough attention: concentrated off-balance-sheet debt among a handful of major tech firms that are essentially lending credit to each other. That is the kind of circular arrangement that looks fine on a spreadsheet right up until it does not.

The stress is already showing up in the market's plumbing. Neo-cloud providers, specifically the repurposed crypto miners now selling compute to AI labs (yes, that is a real business model), are seeing bond yields climb to around 15%. When your bonds yield as much as a payday loan, the market is telling you something. $CRWV (CoreWeave) sits squarely in this category, and that yield signal is not subtle.

Meanwhile, the broader tech selloff is giving this debate real teeth. The NASDAQ 100 dropped 3.5% in October, the Philadelphia Semiconductor Index fell over 7%, and South Korea's KOSPI shed 16% for the month. Nearly $5 trillion in NASDAQ 100 market cap has evaporated since October 6, per jeremiah babe's breakdown, with roughly $2 trillion gone in a single Friday session. That is not a rotation. That is a reckoning.

Where are the defensive plays hiding? Clothier's Capital Gearing Trust is positioned cautiously, and Jody Johnson at Capital Group specifically called out healthcare as both an antidote to AI bubble risk and a standalone opportunity. When two Bloomberg guests in the same week independently land on the same defensive trade, that is worth writing down.

The AI bull case still depends on adoption curves that, so far, exist mostly in pitch decks. When the math requires behavior change at a scale humanity has never delivered before, "implausible" might actually be the polite word for it.