So the problem might be two-fold. First, there's an oversupply of companies trying to use AI relative to the current technological capabilities. Second, even when value is created, the companies don't capture the value to profit out of it.

Tens of millions of freeloaders too.

Every single Ad blocker users is a freeloader to Google/Youtube.

Didn't stop them from being extremely successful.

Pretty sure YouTube has been losing a lot of money every year before they very aggressively ramped up ad frequency and duration as well as subscriptions in recent years. They could only do that thanks to Google’s main cash cow; YouTube would have been dead if not acquired and subsidized by Google.

I am both a Youtube Family Premium and a dedicated ad-blocker user.

studios are the target here, not consumers. Pareto principle applies when you need more than a passive user. 20% or less of the serious studios (which is already a minority) will end up providing 80% of the value of any given AI solution.

...and thousands of enterprise customers translating into billions of dollars worth of deals.

The point is, OpenAI can afford to have free-loaders as long as their deals from enterprise, governments are paying for the service.

Midjourney doesn't have a free plan so no free-loaders there and they're making $200M+ with no VCs.

Stability.ai will always suffer from free-loaders due to their fully open source AI.

> Stability.ai will always suffer from free-loaders due to their fully open source AI.

Stability's recent models (SD3, SV3D, StableLM2, StableCode, and more) are neither open licensed nor planned for release as open licensed.

Sounds like a fairly precarious position they’re in though. This isn’t some new magic tech anymore.

Stability has dreamstudio.ai, but it never seemed like they were seriously investing in it to try to compete with Midjourney?

Yeah, but not sure if OpenAI is profitable yet (they lost $500M last year), and costs are rising, and competition increasing.

GPT-4 cost $100M+ to train (Altman), but Dario Amodei has said next-gen models may cost $1B to train, and $10B models are not inconceivable.

I'd guess OpenAI's payroll is probably $0.5B (770 highly paid employees + benefits, not to mention hundreds of contractors creating data sets).

It would be negligent for them to aim for profitability right now.

They're doing what they should: growing the customer base while continuing to work on the next generation of the core technology, and developing the support code to apply what they have to as broad a cross-section of problems as they have the potential to offer a solution for.