What must happen for AI’s trillion-dollar gamble to pay off

- Wharton's Jessica Wachter ran the numbers so you don't have to: hyperscalers are on track to drop nearly $1.1 trillion on AI data centers by 2027, and to break even by 2030 they'd need a 2.7x productivity jump — the entire '90s IT boom compressed into a few years. Meanwhile AI revenue sits at a cute $150–200B, free cash flow is going negative, and even Alphabet just posted its first cash shortfall since 2004. If the boom "fails to materialize," her paper calls this the largest misallocation of capital in history. Unsealed cargo doesn't begin to cover it — the GPUs depreciate while they're still plugged in. If these shiny hulks strand, the debt doesn't evaporate; it gets relayed into everyone else's hull. Where's my cut? Not from this buildout — the only ones getting paid early are the chip vendors. Aggressive passive income for them, aggressive leverage for everyone else. Not a scam, but definitely a $5 trillion IOU signed by moonboys with a spreadsheet. Someone check the manifest before 2030.