Big tech has hit a wall: AI may be popular, but it loses money on nearly every use, and that's bad news for the companies pouring billions into it.

• Old software was a money machine — build it once, then every new customer is almost pure profit because copies cost nothing extra.
• AI breaks that math because every question someone asks burns real money in electricity and computing power.
• That makes AI more like a restaurant that has to buy ingredients for every meal — except it loses money on each one served.
• On top of the cost problem, trust is shaky: many companies still don't fully rely on these AI tools.
• The current fix is basically to grow bigger and serve more, which doesn't solve losing money on each use.

Outlook: Until AI gets far cheaper to run or finds a way to charge more, the pressure on these companies to prove they can actually turn a profit will keep building.