Big tech's AI spending spree is colliding with power shortages, aging chips, and hidden debt, and the warning signs are flashing for investors and ordinary savers alike.

• Data centers worth billions are sitting dark because the power grid cannot supply enough electricity, with wait times for new connections stretching to a decade or more in places like Virginia and Texas.
• Companies like Meta, Microsoft, and Oracle are spending hundreds of billions on chips and buildings while quietly stretching out depreciation schedules and moving debt off their books to hide losses, tactics investor Michael Burry compares to the WorldCom accounting fraud.
• Electricity costs from data centers are already raising household power bills, and much of the AI buildout debt is being funneled through pensions, insurance funds, and 401(k)s, putting ordinary people's retirement savings on the hook if the bubble bursts.
• Nvidia is both selling chips and financing its own customers like OpenAI and Coreweave, making the AI boom's "growth" numbers look stronger than the underlying business reality.
• OpenAI itself is burning billions more than it earns, relies on non-cash Microsoft cloud credits instead of real cash, and may ultimately need to be absorbed by Microsoft; meanwhile companies that fired workers for AI are starting to rehire them as the technology fails to deliver.

Outlook: Unless power supply catches up or AI profits suddenly materialize, analysts expect a painful correction that could hit tech stocks, retirement accounts, and the broader economy, echoing past bubbles like the dot-com crash.