Comment by mike_hearn 2 days ago
We don't actually know how much money DeepSeek spent or how much compute they used. The numbers being thrown around are suspect, the paper they published didn't reveal the costs of all models nor the R&D cost it took to develop them.
In any AI R&D operation the bulk of the compute goes on doing experiments, not on the final training run for whatever models they choose to make available.
What makes you say that? Do you think Chinese top tier talent is cheap?
I did not refer to the talent directly contributing to the technical progress.
P.S. - clarification: I mean not referring to talent at OpenAI. And yes I have very little doubt talent at DeepSeek is a lot cheaper than the things I listed above for OpenAI. I would be interested in a breakdown of the cost of OpenAI and seeing if even their technical talent costs more than the things I mentioned.
Do you think 1.5M a year compensation is cheap? That’s in the range of OpenAI offers.
What is cheap? But compared to the US, yes. Almost everywhere talent is 'cheap' compared to the US unless they move to the US.
How experienced are you with Chinese AI talent compensation?
I'm sure the salaries at Deepseek in China were lower than the salaries at OpenAI.
A qualified guess. Do you have something that indicates dev salaries are lower in US vs China?
How much cheaper? I’m curious because I’ve seen the offers that Chinese tech companies pay and it’s in the millions for the top talent.
One thing I (intuitively) don't doubt - that they spent less money for developing R1 than OpenAI spent on marketing, lobbying and management compensation.