I was hunched over in my study helping a client grind out a side project when I saw this FT headline. Oracle signed a contract that basically says "pay up whether or not there's electricity"—meaning even if the local grid craps out and power can't reach the data center, they still owe billions for that reserved capacity. On the surface, this looks like big-company contract games. But for people running side hustles, building personal brands, or running tiny teams like mine, it's actually a signal worth chewing on.
Strip it down: AI model training and inference are insanely compute-hungry, and data centers worldwide are scrambling to build and lock in long-term deals. Oracle, Microsoft, Google, Nvidia—they're all fighting for land, power, and cooling. Oracle stepped on this landmine because when they signed PPAs (power purchase agreements), they absorbed all the risks themselves—grid delays, permitting holdups, the whole mess. Who's paying attention? Investors, infrastructure folks, AI application startup teams, and anyone wondering "is the AI bubble real?"—they all have to wade into this conversation.
We don't need to copy Oracle's playbook. What we should copy is the move of "reading industry signals." Barrier to entry: zero—skim the FT Chinese summary, read a few industry takes, you're done. About 30 minutes. Time commitment: spend an hour a week tracking "AI infrastructure" news. You don't need to be technical. First step: treat this as an "industry thermometer"—if even Oracle is locked into contracts like this, it means this AI infrastructure wave is real hot, and also burning serious cash. You don't have to go all in, but keep "which way the tide's flowing" in your head.
If you're a pure side hustler (running a WeChat blog, doing Xiaohongshu): don't stress, but you can use "Is the AI bubble about to burst?" as content material—it's naturally clicky. If you're a 1-5 person team picking SaaS or AI tools: watch whether the services you rely on are burning big money behind the scenes (Cursor, Midjourney, etc.). If they suddenly hike prices or shut down, you need a backup plan ready. If you're building on the AI application layer (calling APIs to assemble products): remember, infrastructure costs always trickle down to the app layer. When you calculate unit economics, factor in "upstream can hike prices anytime"—otherwise a surprise bill will hit you sideways.