Plot twist: this article was written by ChatGPT. /s
Just bet on MSFT and Google.
I would say right-sie. There will be a shaking and the unprepared might be driven out of the market. It just won’t disappear.
GPT summary for Freeper day drinkers:
The article argues that the enormous surge in AI infrastructure spending **probably is not yet a bubble about to burst**, comparing today’s investment cycle with historical booms in railroads, electrification, telecommunications, and housing; it introduces a “Rule of 25,” suggesting serious danger may arise when cumulative investment reaches roughly 25% of GDP, while current AI spending remains well below that level. The author says the bigger risk is **how future AI construction will be financed**, as companies increasingly rely on debt, private credit, equity, leases, and joint ventures, making the sector more vulnerable to rising interest rates and tighter credit. At the same time, strong demand for AI computing and growing revenue could sustain the buildout for years, although weaker companies may fail while major infrastructure providers benefit.
Earlier massive investments were in hard infrastructure that had 25 to 35 year lifetimes and similar obsolescence cycles. That justified huge investments with long payback times. Modern data centers may become obsolete and worthless long before they have reached simple payback. That changes the financial calculus enormously but nobody seems to appreciate that. High debt/equity ratios and short economic lifetimes is a recipe for disaster