^^^
They aren’t trying to be profitable. Their main goals are getting investors to invest even more into them by hyping up their technology as well as building up their market share. Most people are only using the free tier of their AIs which costs them billions, but they don’t care. Profitability can come at some unspecified later date and they probably don’t even have concrete plans for that right now.
I don’t really think they care if it’s profitable. If they can replace labor with something that wont ever rebel or push back they will have even greater control over normal people.
They spend more than their revenue.
and their revenue is miniscule, because they’re afraid to charge the actual costs as that would make people less happy with the lying/hallucinating garbage output they create.
when you’re only paying a fraction of the cost, or none at all, the occasional oopsie doesn’t upset. when you’re paying premium prices for hallucinated gibberish the thing made up, it stings lol
So, a number of companies here in the US, especially in the tech world, and especially B2C, have low variable costs and high fixed costs. That is, it costs them very little to service an additional customer (usually just some extra server time), but they have to pay a lot of fixed costs (things like software engineers to write software) that don’t change regardless of how many customers they have.
If you are a company in an economic situation like that, it is extremely bad to be small, because you are paying those high fixed costs without revenue from a large customer base. That means that it is absolutely vital to grow as quickly as possible, to get out of the “being small” stage. It’s imperative to expand your userbase as quickly as possible.
An added factor is that a number of companies — like those doing social media — work in an area where network effect is a factor. There, the value of the service to existing customers that the company provides rises as the userbase expands. The total value of the service to all customers is something like the square of the size of the userbase. For companies like this, becoming large is even more important.
So what a number of companies in this area have done is to get a lot of capital from investors and then run a “growth phase”, during which they accept very large losses to grow quickly for as long as they can get investment capital to keep growing. They don’t worry about making money much or at all during this phase — they just want to be as appealing as possible, to get as many users as possible, and get out of the “being small” phase. They cover the losses with investment dollars; investors understand that this is part of what they’re signing up for. Then, later, the companies have a “monetization phase” where they worry about being profitable. Usually, that phase has the companies doing things that users like less (since not doing things that might deter users from signing up was one thing that they did to grow quickly).
Cory Doctorow coined the phrase enshittification for the transition between the two phases; the user experience in the monetization phase tends to be worse in some ways than during the growth phase.
AI companies are all pretty young (at least, as regards their AI aspects; some are existing companies moving into the space), and are in the growth phase now.
are in the growth phase now
Speedrunning late stage enshittification, really.
yeah that’s a good explanation why there’s only a very small number of software companies in the world. google, microsoft, apple, meta. the reason is because, when you have two cars, that’s twice as much as one car. but when you have 2 apps, that’s worth exactly as much as having 1 app.
consider this: scenario 1: one big company writes one calendar app that everybody uses. scenario 2: there’s two medium-sized companies writing calendar app, that share the users. Which is better?
two companies -> twice the fixed cost (writing the code twice for no reason). two database protocols -> incompatibilities, so users sharing data with each other becomes more difficult, for example for group calendars where events are distributed to the app that the user already has. this is also called “network effects”: removing boundaries by everything being on one platform.
downsides of monopolies: one company might have too much market determining force. no competition, therefore difficult to evaluate what would happen if things were done differently.
that’s why there’s no second search giant besides google. for mobile and desktop operating systems there’s two, probably to have some competition (android/iOS, windows/macOS).
meanwhile there are no such monopolies for car companies, because if you build twice as many cars, then you have twice as many cars. so competition pays off.
There are many more software companies in the world then those four, including very small ones. It is still possible to make a reasonable living as such a small software company, though a lot harder than it used to be.
To be clear, so-called AI isn’t at all about improving anything for anyone other than the ultra-rich, who are propping up this loss leader. They are using it to control you, to spy on you, and to keep you in your place. That is it. That’s the entire story.
They are deliberately operating at a large loss to race ahead in capabilities. There’s no second places in this race. Whether it’ll actually pay off in the end remains to be seen.
There is a difference between AI and LLM.
There are AI programs that make a lot of money. These are generally bespoke AI programs designed for a narrow set of tasks, like detecting tumors from an MRI scan.
Most LLM’s and image generators don’t make money because computing cost is significantly higher than the value of the output.
The infrastructure required is a huge investment which has to be recouped through monthly/yearly subscriptions.
AI companies aren’t profitable because
- nobody really wants to use it. we all know it’s a walled garden. OpenAI is gonna enshittify just like google and microsoft did. never put your infrastructure into another company’s hands. that’s a recipe for making yourself vulnerable and you’re gonna dearly regret it later. at this point, trusting an US company with your data is a typical example of insanity. In europe, practically every big company/government institution is trying to get away from the dependency on US tech, not towards it. As long as AI is all hosted on US company server, nobody’s gonna use it. It would have to become self-hosted and open-source/open-weight before that.
Oh no. People want to use it alright.
Because its all built on lies
Its what happens when you have a good demo but can’t make good on the promises you made in that demo and have no viable monetization plan.
And with help from AI, I could soon have the exact same problem!
They have watched too many movies.
And in the movies there is 1 winner in the end who has the best AI and then he owns ALL THE POWER.
Now they all want THAT. Profit comes later.
They dont need to be profitable.
Because each prompt is extremely expensive to generate. So expensive that nobody would want to pay for it.
However, the techbros want everyone to use the AI hoping it reaches the point where it becomes profitable so what are they doing? Eating the costs.
There’s no time for profits. AI investors don’t want boring old companies that make tools which do useful things and earn revenue — they want maximum investment into the most expensive AI techniques possible where they feel like there’s a chance that all their money might cause AI to reach critical mass and then shower them with riches beyond what pre-AI people can even imagine.
