This is happening though. The AI foundries are rapidly expanding into new businesses.

I opened ChatGPT and today it offered me “Finances” which would allow me to connect my financial data with ChatGPT and it could help me manage my subscriptions, etc.

Right now they want to improve their models. That means still getting a lot of data from around the world. Companies wouldn’t give them that data if they thought they might take it and put them out of business.

The smart thing to do would indeed be to pretend to be a service provider until you have enough information that you don’t really need 3rd party data anymore, and then to start putting other companies out of business.

I suspect these companies are less than 1-2 years away from creating a DoorDash/Uber Eats competitor product.

If there were only one firm, then it would make sense to not sell inference. However OpenAI, Anthropic, Google are all in competition with each other to earn market share, developer goodwill, and to get reliable growth, selling inference is the best strategy, rather than making a thousand companies and needing to do the marketing, branding, differentiation, PR, etc for each one seperately.

Also, the AI companies learn a lot from selling their inference. They see what markets are growing fastest, what people are using AI for, information benefitted by the startups using their models putting way more effort into distribution than they are. With enough information, the model companies can then create a generic AI-enabled product (like they have over the past year) that factors in all the learnings the startups have undergone.

I think AI is more like the fundamental layer kind of changing, that reshape the values, rules and perspectives in modern society. Examples: inventions of paper, gunpowder, train, airplane, internet, and now AI. No body can claim the ownership of this change, and everybody can participate in some ways, selling it, using it etc.

AI hype is real, and there is a lot of perverse incentive to keep the hype cycle going. You are right about that.

Rest of your analogy/arguments are just a lot of nonsense too.

Electricity is valuable, and by that logic Siemens Energy should build and own all factories under the sun. TI should stop selling their calculators and compete in Math Olympiad instead!

More like a super expensive version of Factorio for engineers. Everybody I know has a software factory now, and a whole bunch of software nobody wants :)

The only person getting rich is a Taiwanese businessman in leather jacket.

You still need solid motivation, a good sense of the fundamentals, communication skills, social skills, sales skills, and good taste to build good products with AI.

Also, I think you are conflating two things:

1. The genuine advancement (and challenges) AI brings.

2. The inevitable hordes of less scrupulous people who take advantage of every new trend to make money however they can.

I am fully immersed in building with AI and I barely notice these less scrupulous. I know they exist. They always exist. But why are you paying so much attention to them?

Are shovels just another get rich quick scheme? If they were really all that, they wouldn't be selling shovels to the public. They would be forming corporate partnerships, digging dirt and conquering markets--starting with the mining industry.

the ai companies have a problem at the moment: the compute required for inference (especially) costs a metric shit-ton of money - so they're crowd-sourcing the funds (unsuccessfully so far, tbh - for example, a $20/m claude sub that i'm on for work, and I regularly see individual conversations where the tokens cost more than that - even up to $350.

It's a tool tho, not a panacea. With the right guidance, for example, you can get some good code results.