The title is an exaggeration.

OpenAI is a systemic risk only to current tech valuations and to the near-term availability of fresh capital for new AI infrastructure.

Whatever happens to OpenAI, to valuations, and to the availability of capital over the short run, technology will continue to progress over the long run, in fits and starts, as always.

I've liked some of Ed's previous writing, but this is a craaaazy statement: "The Future of Generative AI Rests On OpenAI".

OpenAI is an over-hyped, over-priced, and under-performing AI company. Fortunately, the underlying LLM/transformer technology is real and valuable and not at all dependent on OpenAI. It would be good to deflate some of the hype around OpenAI and other non-viable companies.

It's unclear how OpenAI intends to quadruple its revenue? They just doubled their active user count in only a few weeks and their Pro subscription was a new offering as of January. I'd be surprised if their revenue wasn't at least four times larger by the end of this year.

Are there any precedents for a subscription service that went from 0 to $5 billion in revenue in two years? I also think, they are doing a ton of expensive R&D, but as far as I can tell that $5 billion in revenue is a profitable and sustainable business. It's not like they're selling compute below cost.

They went from 0 to $5 billion when they were the only game in town. But now the crown of "best model" changes hands every few weeks, from OpenAI to Anthropic to DeepSeek to Google, and that looks likely to be the case for the foreseeable future. People "in the know" have already stopped treating OpenAI as automatically the best, and this will leak out over time. Without the best model as a moat anymore, there's no reason to expect OpenAI's growth to continue on the trajectory it used to have.

Uh, they are selling compute below cost. They are losing money on every subscription.

I’d be more interested in user retention. Every AI product I’ve seen (from the investment side) does not speak of this. I have seen some data though that suggests user retention is short (3mo avg). People try it, find it’s not that useful or even harmful after the initial experimentation and dump it.

Gotta keep hype to keep MRR up though even if it’s from different people. They will run out of interest and new users soon. Going to be a big fall.

Models will stagnate on this funding crush and the promises will be gone in a puff of smoke. And everyone will have to unfuck their dependency on it for upselling their existing crap to end users.

The author previously wrote an article about dissecting this very question:

https://www.wheresyoured.at/wheres-the-money/

What's interesting is that they stopped reporting monthly active users and prefer weekly active users. The author argues that this is a bad thing, but I'm having trouble understanding why?

Great article! It is hard to see an ROI on OpenAI at this stage, or indeed a sustainable business model based on revenue rather than investment.

That does not capture API usage. I use Claude every day but never through the app. It is very far from crappy. It is top tier.

I don't think they are differentiating between types of users, just amount of users.

You can't count the amount of users by looking at app usage; it's not representative.

My two cents of common sense would suggest that API usage looks very similar.

How would they look similar? I just told you I don't use the app at all, not 50%/50%? Anthropic targets the enterprise, while OpenAI targets the consumer. Anthropic's primary product today is the API.

By similar, I mean that OpenAI API usage should also be disproportionately higher than Anthropic's.

But you're free to prove me wrong with some data!

Show me the data.

The article you linked says that 60-75% of Anthropic's revenue comes from API calls (which would include things like Cursor).