The entire US economy is currently being propped up by growth in the AI/tech sector. And I am convinced that LLMs are fundamentally incapable of delivering on the promises being made by the AI CEOs. That means there is a massive bubble that will eventually burst, probably taking the whole US economy with it.

Let’s say, for sake of argument, that I am a typical American. I work a job for a wage, but I’m mostly living paycheck to paycheck. I have maybe a little savings, and a retirement account with a little bit in it, but certainly not enough that I can retire anytime in the near future.

To what extent is it possible for someone like me, who doesn’t buy into the AI hype, to insulate themselves from the negative impact of the eventual collapse?

  • humanspiral@lemmy.ca
    23·
    17 hours ago

    Weird question. Not clear anything you can do.

    First, AI bubble means datacenter bubble. Nvidida, AMD, TSMC, Chinese equivalents will do fine, as they have options to make products for non datacenter use.

    Scenarios:

    1. No mass corporate uptake for datacenters, or requirement to encrypt upload/download traffic to corporate owned models hosted by datacenters. Amazon/Google/MSFT can win relatively such a race if they allow private encrypted models instead of their own, and can buy distressed assets from failures. It just means slower than announced deployment rates, with only losers those datacenters who get married to losers.

    2. CPU enhanced AI (knowledge graphs) with/or smaller LLMs. Datacenters can still provide corporate users, but mix of regular and gpu datacenters, Datacenters can lose big if next big thing requires replacing hardware, and they were too early. Shift in winners and losers, but not an AI/datacenter bubble.

    3. Few of the announced datacenters are ever built, or 5 year+ delays. Public company investments will go down a little, but nothing catastrophic for big tech, who can make it up in other areas. Power company histeria is an associated bubble that does poorly. This is a very likely scenario.

    4. Datacenters are successful and aggressively built. No AI bubble, because government surveillance revenue is obtained, and heavy government use of LLMs to keep population pro Israel/oligarchy/militarism. A freedom and jobs bubble is not better than an AI bubble.

    Meme stock mania means even the biggest losers can rebound strongly. An everyone else bubble happens whether or not AI datacenters are successful.