Thinking in Public
Software sells tools to workers. AI sells the work itself, the work people do on that software. That is why AI is so big and valued in the trillions.
The world's workers get paid about $60 trillion a year in wages, and AI is vying for all of that. The ILO says workers earn about 52 percent of world income, and the World Bank puts world GDP at about $118 trillion for 2025. That makes about $62 trillion.
It is potentially worth trillions because it will do so much of the world's work. So it is valued as the world's work and not as software, which is crazy.
Every single industry is a customer, including governments.
Part of the valuation is a bet that it will build something smarter than workers. Smarter than us. So it does more, at more scale.
Nvidia, which sells the chips that run it, is valued at about $5.3 trillion as of October 27, 2026 (CompaniesMarketCap). Those are the tools. The models are next on this list.
It's already happening. Coding is mostly AI: Google says 75 percent of its new code is now AI-generated and approved by engineers (Google). Driving: if you've been to California you can see it in action, with Waymo running driverless taxis in Los Angeles, San Diego and the San Francisco Bay Area (Waymo). Tutors and call centers too, and the next things are legal and accounting.
It starts here, with tasks like translations.
It is slowest where liability and regulation are involved.
Slow too. Once these models have hands, they'll be able to do anything.
It's not happening tomorrow, but that is the direction, and it makes AI a huge market.