Hacker News Reader

Comment by user43928 | Hacker News Reader

> All this must mean the Western AI companies are now extremely inference-margin positive.

> So the Chinese labs have thrown a lifeline to the Western loss-making labs, and I just have no clue as to why.

That inference wasn't profitable is a widespread myth.

Analysis based on Kimi K3 suggests that OpenAI and Anthropic have margins well north of 95%: https://inferencex.semianalysis.com/run/kimi-k3-on-b200

Over the last months I have seen news that OpenAI made breakthroughs in inference efficiency multiple times.

I have no reason to believe that the leading US labs don't have their own optimizations, or that they learned of this particular optimization from DeepSeek.

View on HN

> I have no reason to believe that the leading US labs don't have their own optimizations, or that they learned of this particular optimization from DeepSeek.

If they already did, then DeepSeek still made them discount their prices significantly, which eats margin.

Yes, competition is great for us.

I wonder if margins on GPT-6.1 Sol and Opus 5.5 are now 75% or 90%.

> 75% or 90%

The difference matters when they're investing the excess into salaries and bonuses to build the next frontier model.

Seen from a high-level perspective, if Chinese open models are compressing US AI labs' profit margins and those margins fund US AI labs' dominance, then open models are decreasing American AI dominance.

95% margin is really unlikely. Anthropic recently said they have 80% gross margin when using their adjusted ebidta (ie if they do not consider revenue sharing, training expenses, and a bunch of other costs). They wouldn’t be talking about non standard metrics if they had such high margin on inference

What we were talking about here is the margin on inference as in:

Cost per GPU hour versus API price of generated tokens assuming 100% utilization.

This could be a margin around 98.3% for 5.6 Sol.

If the utilization of the GPU was 25%, it would drop to 93.1%.

Revenue sharing or training expenses are not considered here in this "inference margin".

Reply on news.ycombinator.com