supercat
Major
On a PPP basis, China's public budget for R&D is more than America's now. I think most of those who works for basic research in the US is foreign graduates.
News on China's scientific and technological development.
- Thread starter Quickie
- Start date
PPP is not a good metric here. PPP calculation is done by using a basket of goods, for research that basket is extremely different:
- Loads of funding for research in science/tech would go in infrastructure and equipment. That is probably similarly priced (maybe even more expensive in some cases) in China. You can't buy top end GPUs using PPP right? Similar for a lot of other things.
- Top tier talent is globally mobile meaning that their salaries, while lower in China, wouldn't be as different as suggested by PPP.
- These numbers are intended for western audiences to push for more funding in US/west to counter China. It's deliberately playing up the Chinese figures.
Take the Fraunhofer Institute, who are world leaders in terms of applied research.
If you look at their financials, they spend around 100K euro per person.
Average salary looks around 50K. Add then another 20% for indirect labour costs.
You can already see that a slight majority of all spending is on labour.
===
On infrastructure and equipment, I would say that domestic Chinese equipment does have a significant overall advantage in terms of costs, when compared to elsewhere.
On the topic of AI, Macro Polo reports that in 2022, half of the world's top-tier AI researchers did their undergraduate degrees in China. We can expect this trend to continue in China's favour, given the AI frenzy in China, which is even more pronounced than in the US.
Many Chinese researchers work do not bother with leaving China, given the language and culture differences.
And of the Chinese researchers in the USA, given the hostile political climate, they will be subject to a glass ceiling. That will drive them back to China where they have better opportunities.
On AI hardware, we see an arms race, to build every larger Language Models. But I'm doubtful how useful this will be, because the results all depend on accuracy of the data and also the viewpoint fed into the models.
For example, a model trained on US data would support Israel, and omit the fact the successive Israeli governments and the Israeli Army over the past 50 years have supported what is now 700,000 Jewish colonists in trying to take the West Bank from the existing Palestinian inhabitants.
In comparison, models trained on data from the vast majority of countries in the world would take the viewpoint that Israel didn't exist a hundred years ago, and that Israel is a colonial construct that owes its existence to taking land from the existing Palestinian inhabitants, and which they continue doing even today.
So that leaves more specialised machine-leaning models, which do not require as many resources. Ideally, you want to create as small a model as practically possible, so that it is more cost-efficient (and commercially competitive) than a larger model.
===
Top-tier talent is not globally mobile, because China is very different in terms of language and culture. It's a lot easier to move around if everyone uses English.
China has a lot of advantages in this:
That's $1.5 billion nominal, make it PPP, and that's a lot more
Sour grapes.
So China is (as usual) actually following a long-range plan while the US just talks about it (as usual)
China's real strength is in scaling. Even if China was a bit slow in R&D, when it would go for scaling, fusion energy plants would be built in sausage-mode in the entire country which would completely power up Chinese industries, R&D and national comprehensive strength. It would a be a complete and utter rocket boost for China's capabilities
As they always say bro, behind every Chinese success is a White man.