China needs mechanisms that legitimize and absorb failure at scale, while preserving incentives for extreme upside. Otherwise, the system collapses into path dependence: capital chases what it can already model, and anything truly novel never even enters the decision space. That is a structural blindspot.
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
China needs mechanisms that legitimize and absorb failure at scale, while preserving incentives for extreme upside. Otherwise, the system collapses into path dependence: capital chases what it can already model, and anything truly novel never even enters the decision space. That is a structural blindspot.
You might have a point. However Chinas system is different ( every system has its advantage and disadvantages). You can’t have the best of both worlds . lol. China system is still more cautious and prudent about investing huge sum of capital and funds into unproven technology , which is rational . Don’t forget that China is still technically a developing country . I think Chinas rise as a tech power make people forget this fact. China is not yet a fully developed country unlike the West, Japan, US etc. So Chinese leaders still need to also prioritize lifting the remaining people over the edge of poverty , especially those in remote areas . This is a command able task Xi Jinping has set for the government which is a very commendable and rare nowadays . Instead of putting tens or hundreds of billions into some project that nobody knows how it’s going to end and many suspect it might be a bubble (it’s possible it turns out it’s not as well).
so we need to consider the reality of Chinas situation first . So I’m actually in support of Chinas piece meal a cautious method . Better to wait and see while investing more into application of what is already there and scale it . It’s not too late to catch up if the Americans take the risk and make it work . lol. China is god at that . Once the country is fully developed then the government can afford to spend such sums on projects like this
What are they in denial of?
Bytedance, Alibaba, Tencent, Baidu, etc, are all investing massively in AI. This isn't like SpaceX. If you really want an analogy, this is far more like when Tesla built a factory in Shanghai. Well, where is the EV industry now in USA and China?
Oh yeah SpaceX Elon Musk little darling, one of the most subsidize companies in the US (nothing wrong with that but), the company that basically depend on starklink satellites, Musk another company.
And no, Anthropic is not growing 10X every year in revenue if you believe their numbers, they are growing from their startup base to more stable revenue, but that I think doesn't worth a trillion dollars that is no even logical. Is like saying that Ziphu grow from zero to a billion and now they are worth 200 billion, that is illogical, that is bubble that put a lot of pressure in the company to deliver but when reality comes the bubble will pop.
They are facing grow pressure from all sides, from other US companies like OpenAI and Gemini (really dangerous) and Chinese companies, they face increasing costs and costumer's face increasing pricing and with users now dividing their workloads to cheaper models like DeepSeek. That put more pressure in them than to Gemini for example who parent company depends on ads and other services revenue, that is why Anthropic is that biggest advocate of regulatory capture.
I don't hate Anthropic, I don't care about them, but I don't hate them either, they offer a good service (no a product) but I think is the poster child example that this AI Mania is reaching crazy levels.
What do you mean Chinese analysts were in denial? Isn’t Manus one of the first — or even the first — agentic AI systems to actually function the way agentic AI has been marketed and performs today?
Where China may be lagging is in the slower adoption of tools like “Claude Code” from Anthropic or “Codex” from OpenAI.
And just to add to the list of concerns that many American AI firms are going to be facing or must grapple now is from the following:
Especially the following line: A senior official from a US-friendly government told me: “We will not use your solution if it relies on a US-based LLM.” It wasn’t a joke.
The US is THE financial super power that can throw capital around like it's nothing because of dollar hegemony. As long as dollar hegemony continues, it will always have an investment edge over other countries. This is the reason regions like Europe and East Asia have gone in the directions that they did. It's not because they are incapable of innovation, it's because the capital environment outside of the US simply cannot match up to the US, which can print trillions of dollars and get the rest of the world to subsidize them for it.
Why do you think the US stock market is >60% of the global equities market? It's because everyone invests their money here, because you are able to get minimum ~10% returns on investment each year over the past 40-50 years, which no other equities market offers you. Even though China dominates global manufacturing and has a much stronger industrial chain, it remains the case the US has much higher return on investment, so that's where all the investors go.
With such a large share of global capital, R&D projects for the US are effectively without risk. Trillions of dollars can be wasted (and are wasted) each year on projects that go no where, but the few projects that do succeed (SpaceX, AI, etc.), have global impact. China - and the rest of the developed world - plays "fast follower" for a reason - it's the only real path available to countries that do not control global capital.
A descriptive of Chinese companies strategy with Open Weights
TLDR: They want the ecosystem and they want to sell solutions no just API Calls.
TLDR: They want the ecosystem and they want to sell solutions no just API Calls.
Its funny you bring up SpaceX, because its actually a perfect analogy for Americans living in denial about their inability to compete using brute force against much cheaper options built on better tech. Keep in mind ULA's revenue was higher than SpaceX for most of its history too.
And personally I find Kimi's quality to be better than Claude in enough cases to go 100% Kimi, and this is coming from a $200/mo Claude tier that I didnt have to pay myself. If I did the price difference would make it an even bigger nobrainer
Claude is not bad by any means and I perfectly understand why people use it extensively, but it simply cant justify its pricing when at best it matches Kimi at 5x the price (or infinte if you self host), or in my experience gives worse outputs.
IMO people who pay for Claude are just paying for the brand and dophamine induced mental association, or just because their company pays for it. Anthropic is a business development and political capture play, not capability or tech.
Its wine tasting and the wine you find taste worse wants you to pay 5x more. There will be people who swear by the 5x wine just because they paid 5x and dont want to think themselves fools, but objectively its not worth 5x and they are fools
Last edited:
Is just basically speculative money backed by usually debt and boomers pension money.