
@quxiaoyin You even appeared on Singapore's businesstimes 💪🏻
https://www.businesstimes.com.sg/international/ai-linked-layoffs-us-spark-new-wave-one-person-start-ups
Carlos Perez responds that revenue maximization is not the sole driver.
Many users praised Chinese open-source AI models and engineers for advancing the field, while others dismissed the models as low-quality or risky due to government ties.
No Digg Deeper questions have been answered for this story yet.

@quxiaoyin You even appeared on Singapore's businesstimes 💪🏻
https://www.businesstimes.com.sg/international/ai-linked-layoffs-us-spark-new-wave-one-person-start-ups

Frontier models have indeed become the shittiest, most commoditized layer, now forced to slug it out with incredible open-source alternatives that deliver massive value for free. OpenAI and Anthropic built something amazing, but to justify those sky-high valuations they’ll need to aggressively expand into compute, energy, vertical applications, and full-stack solutions where real moats and margins live.
China has completely game-changed the AI landscape for the better, accelerating commoditization, driving down costs, and pushing everyone to innovate faster. The rising tide is lifting the whole ecosystem.

@tudorvafe Not true. People are lazy and not pivoting that fast especially if ai has all the context.

@quxiaoyin Anybody using CN LLM is delusional! ITS ROUTED THROUGH CHINA GOV!

@quxiaoyin Anthropic has grown revenue faster then any other company in history. That qualifies as ‘not making money’ to you?

@nick_archuletas I love open source. It’s so good now

@quxiaoyin compute and energy, agreed — those are physical and scarce. but 'applications' isn't a moat, it's just the next layer to commoditize once the tokens underneath are free. the durable rent is the part you can't spin up with software.

@quxiaoyin Intelligence per buck

@RayT168 lol I did

@quxiaoyin A breath of fresh, based air on the timeline. Are you having fun running any of the crazy new open source stuff on your own silicon yet as well?

@noctus91 Not true anymore.

@quxiaoyin The Chinese models are really not good at all, they work well for simple engineering tasks I can't use them too much stuff broken, they are also super lazy and burn a shit ton of tokens to produce garbage.

@quxiaoyin extremely true

@quxiaoyin your posts are changing policy notice how they brought back fable5 lickety split 😁

@quxiaoyin But isn't the model itself the product?

@quxiaoyin I can't believe that at some point we were discussing that future of software engineering is not secure bcz of AI
Now the future of those companies itself is not secure 😀

@quxiaoyin Hm, except they didn't invent anything. LLM isn't new. Neither are transformers.

@quxiaoyin I fucking love China even though I still use Gemini, switching soon.

@quxiaoyin @quxiaoyin
This is why i've been on Eastern models for months- High performance- reasonable costs. I'm running DS,http://Z.Ai, Minimax and Moonshot. Between GLM 5.2 and DS/M3 I'm shipping and not drowning in costs. Very pleased.

@s_batzoglou Awesome metric. Agreed