[ansk]

This dude very well may be better off with the 100 bucks at this point. The company is bleeding money at a ridiculous rate, so they're going to need to fundraise shortly. Given how little they have to show for their initial funding, they're going to struggle to attract any further investments. The initial investors already look like absolute clowns for giving a $1B valuation to a company whose primary selling points were 1) a model which was state-of-the-art for maybe a month 2) a github repo with a lot of stars and 3) a loose association to some mediocre researchers. Unless they can find someone to hold the bag in the very near future, their equity is going to be just about worthless.

[Filligree]

Let's be nice; they've released Stable Diffusion and now SDXL for free, which has completely changed the landscape on what can practically be done by individuals.

Having an open foundation model for image-generation is a service to the world. It just isn't exactly obvious how it could possibly lead to profit.

[ivalm]

The loose association was with decent researchers, no need to bring them down. I do agree about the rest.

[ansk]

Mediocre in regard to the standard set by other research teams pulling $1B+ valuations. The OpenAI founding team and the recently announced team at xAI are what I would consider to be exceptionally strong groups. Adept, Anthropic, Inflection, and character.ai are also strong. Look at the track records and accolades of researchers at these groups. Ilya Sutskever alone has more research experience than the combined team of phd students Emad was working with (and eventually hired).

[anonylizard]

Wow, the AI field has barely boomed for like 1 year, and we are already all-in on credentialism?

Stable Diffusion has had a very large impact on the entire arts industry, far more people have used SD than have Claude or pi. If another open source startup can reach that level of industry shaking fame and intense user interest, they could have easily be valued at 1B.

If SDXL works out, that'll be another large leap forward. Midjourney's models are stronger, but they can't compete against a legion of 1000 fine-tunes for every purpose.

[Filligree]

Yeah. I still don't see how this will work for _Stability_, but the impact is definitely there.

[ShamelessC]

The researchers responsible for latent diffusion and VQGAN have competed directly with OpenAI for at least 2 years. They are at the top of their field.

[skeaker]

Maybe this is callous, but this kind of sounds like a win win. Investors who are responsible for enshittifying all my favorite stuff finally get screwed for once, AI gets democratized instead of being solely in the hands of OpenAI/Google/etc in the process. Nice.

[EVa5I7bHFq9mnYK]

The dude might well be better off in the near future when the hype passes and company is worth its fair price of $50.