They’re selling hype, not technology
The reality is the product is the easy part.
Actually selling it, maintaining it, selling more of it and running a profitable business is the hard part.
So the AI makes the product really easy to make, but selling it is the hard part.
Why don’t they ask ai how to sell it. After all, isn’t it super intelligent?
I do work for a software company, and sales is part of my job. And yes, sometimes I get the AI to review a tricky sale. Not usually though.
However, the reality is it takes years and years to build relationships, get into markets, get visibility.
Basically embed yourself into a market. So the product is the easy part. The sale is the hard part.
That’s like asking why doesn’t Adobe make their own content and keep the tools internal.
The AI companies could probably be profitable if they weren’t dumping tons of money into new model research, but then they would stagnate and die to the ones who keep moving forward.
Companies using their AI don’t need to worry about that expense.
It’d be so funny if research hits a breakthrough that absolutely collapses the cost of inference, allowing anyone to self host frontier models, eliminating the ability for all the main AI companies to ever recoup their costs.
Well, except where everyone’s data gets sold at bottom dollar in liquidation. Actually, that’d be pretty funny too.
I don’t know if that’ll happen to frontier models, but I definitely see it happening with models that work well enough for the vast majority of personal use.
The real reason isn’t that they can’t turn a profit, it’s that they don’t have to.
They don’t have to be profitable because they aren’t fueling growth with profits, they’re fueling it with debt and investment. Like many have noted, they cannot do that forever and will eventually pass that cost to consumers. A significant portion of their existing cost is driven by research, which they can scale back by lobbying for legislation, so that peers getting a leg up is more difficult and small players catching up is almost impossible.
Also, both Anthropic and OpenAI have planned IPOs, where they will inevitably raise a ton of cash. It’s a game of chicken though. How long can they fuel growth without being publicly accountable to balance sheets and shareholders? Whether we like it or not, they will almost certainly transition to being profitable businesses, but only when it is advantageous for them to do so. Right now, it’s not a priority; the priority is growth to the point of diminishing returns.
To be fair, a reasonable answer would be the same as any reseller or value-added business. They have some skill, capital, or existing market capture that the whole seller does not.
Don’t get me wrong it can still be bullshit, but it’s actually pretty common.
It’s a definite risk. They do steal your inputs and the outputs they send back to you. They will always have even more powerful models that aren’t released yet. If you avoid sending them all of your code, they could hack you to steal it. They will certainly copy any good ideas their competitors come up with (Meta Muse agent, jev) within weeks, as they all copied claude code and open claw. Answer to OP is “just not yet”.
The unified US establishment goal is to create skynet for global and domestic domination. Theoretically, the AI bond villains could make sufficently sophisticated weapon/control systems to threaten the government, but why bother when it already wants to give them all of our money. “Whoever wins AI, wins” means only first to skynet, wins. Zero relevance to any AI customers, who obviously win through the widest competition/choices on what to use to help them. Not the nationality of the oligarchs that profit off them.
https://naturalfinance.blogspot.com/2026/09/a-global-prosperity-manifesto.html
Because its a scam.
A scam that will inevitably end humanity on a global scale…… probably.
Nah mankind will continue, but it’ll definitely hurt when this shitshow implodes and possibly takes out the current economic model and quite a few governments with it.
I’d wager it’s simply not possible to be profitable using AI for most use-cases. You either burn way too many tokens on your queries or risk your business falling victim to an under-tokenized AI hallucination.
If I recall correctly, shovel companies made a lot of money while only a few of their shovels actually struck gold.
Recent scholarship confirms that merchants made far more money than miners during the gold rush.
The wealthiest man in California during the early years of the rush was Samuel Brannan, a tireless self-promoter, shopkeeper and newspaper publisher. Brannan opened the first supply stores in Sacramento, Coloma, and other spots in the goldfields. Just as the rush began, he purchased all the prospecting supplies available in San Francisco and resold them at a substantial profit.
Better question is if you can really build an app as easily as writing a prompt, why would anyone pay you for your app?
Somehow AI bros like to pretend that AI is some secret lifehack despite it being one of the most mainstream things in human history.
They are going for the gold rush pick axe business model
The difference is there was actual gold in them there hills. LLM’s are hogwash
But there is no gold at the end of that rainbow. It’s more like a selling Florida swampland to real estate developers model.
There’s a lot of reasons;
- The people who run these businesses have grown incredibly wealthy by taking investment to grow. They do not need to be profitable
- Selling inference is massively profitable. They lose money on model training
- If you’re suggesting they build a “product” that is profitable, I would argue that’s a bad idea- it’s a high competition space. The people selling AI are in a much better position
Why does it cost so much to engineer new models? I thought AI can replace software engineers now, shouldn’t the development cost of new models decrease over time?
Joking aside - I think the whole “engineer time is expensive, computer time is cheap” approach stopped being true half a decade or so before LLMs were a thing. It was kind of self defeating - you’d think companies were trying to increase their usage of computing resources due to some misconception that it’d automatically decrease work humans have to do, saving them money in the long run.
I don’t have the numbers, but I wouldn’t be surprised if software/hardware costs exceeded engineering costs for about a decade now, with all these inefficient-on-purpose frameworks and with everything being on the cloud and with using various services for the sake of using a service.
That is the right question.
