Hacker News Reader

Comment by atonse | Hacker News Reader

I agree with you. Even Anthropic's CEO said EXACTLY this. He said, if you actually look at the lifecycle of each model as its own business, then they are all very profitable. It's just that while we're making money from Model A, we've started spending 10x on Model B.

View on HN

Exponential cost growth with, at best, linear improvement is not a promising business projection.

Perhaps at some point we'll say "this model is profitable and we're just gonna stick with that".

> Perhaps at some point we'll say "this model is profitable and we're just gonna stick with that".

I don't follow it that closely but my perception is that's already happened. Various flavors of GPT 4 are still current products, just at lower prices.

That is not what GP was saying. They were saying that a foundation model company would say "$CURRENT_MODEL is so good that it is no use training $NEXT_MODEL right now", which I don't think is the current stance of any of those companies.

Given that they are all constantly spending money on R&D for the next model, it does not really matter how long they get to offer some of the older models. The massive R&D spend is still incurred all the time.

Well have they actually come up with data supporting this?

Yes and it’s furnished to some of the worlds best investors who are happy to pay valuations between 150 and 500b to access it

Among those are the guys that gave money to WeWork.

Make of that what you will.

I am not really betting on it.

I don’t quite understand your wording here. Do we have to “pay valuations between 150 and 500b” to access the data that supports the justification of the valuation or can you just link to it?

the logic of a happy WeWork investor

Yeah, no investor has ever been conned before /s

I do hope Ed's wrong, I actually stand to financially benefit from it.

Reply on news.ycombinator.com