I’m curious about the mental model of people who think ai is extremely subsidised. This view is strange because GLM released a few weeks ago and it is confidently better than say GPT 5.

I think you are in for a rude shock - your expectation that there will come a reckoning where people are forced to content real prices of AI.. will never happen. It won’t.

▲

npodbielski 3 days ago | parent | next [-]

And how you do know that GLM is not subsidised? Maybe Chinese government is burning money to destroy US economy by making them lose money in AI companies that are just furnace burning dollars. Unless you have some compelling argument it is just a speculation.

▲

SyneRyder 3 days ago | parent | next [-]

It's because GLM 5.2 is offered on many inference providers, including providers in the US. Those companies only make their money by charging for inference, and yet they seem to be doing quite well while charging the exact same prices as Z.AI / GLM.

In fact, there's a price war where some of the US inference providers are undercutting the pricing of Z.AI's own GLM hosting. Novita & AtlasCloud are both offering 8% and 5% discounts on GLM 5.2 respectively. GMICloud is charging 30% less - but getting so hammered with demand that it only has 80% uptime & 7 tokens per second, so you get what you pay for.

You can find a list of providers & their pricing through OpenRouter here:

https://openrouter.ai/z-ai/glm-5.2#providers

	▲	npodbielski 2 days ago \| parent [-]
		Sure I am not disputing that. But we can't say that z.ai is profitable because they charge the same as other providers per token. There is also training cost. And reaserch cost. Just inference is not all. Maybe there are profitable with all of it, I am not saying they are not, I do not know but at the same time I highly doubt that. Unless somebody will find a way to actually do training by just updating weights with new data you have to train from scratch and this is very costly process. Saying that some lab is profitable because they are profitable doing inference while disregarding training cost is not fair assessment.

▲

faangguyindia 3 days ago | parent | prev [-]

West china has cheap electricity via solar. Satellite images reveal large scale data center build outs. Untill now this electricity did not have much use as this region lacks major connectivity to ports and industrial regions.

▲

blourvim 2 days ago | parent | prev | next [-]

Training is expensive, to keep selling tokens, they have to have the best model. They have to keep spending money on model training

A product or a platform could give them a moat, but they don't quite have that yet although I think they are looking to get there

▲

faangguyindia 3 days ago | parent | prev [-]

It happens because people fail to account for economies of scale. Believing everything scales linearly is major flaw in the ways most people think.