Comment by foo12bar
2 days ago
The Chinese government grossly subsidized many industries to undercut their competitors and form monopolies, such as solar panels, electric cars, battery cells, and steel and aluminum manufacturing, not to mention rare earths. Yes, they are more efficient, I don't disagree with you there.
But they aren't making any money releasing open weight models, and no, I don't believe they are like Linus Torvalds with some grand vision of free as is freedom AI models, but rather doing more of the same.
The US government has also grossly subsidized many industries. Tesla literally wouldn't exist without subsidies. However, the whole premise behind capitalism and markets was that this system was inherently more efficient than state planning. Now we see that narrative is false because China's planned system is winning across the board.
And that's the beauty of a planned system, you don't have to focus solely on short term profit there. If Chinese government sees this tech as foundational, which they do, they can just keep pouring money into it because it provides general benefits for the country.
This is exactly what happened with high speed rail incidentally. People in the west kept talking about how unprofitable it is, while China understood that creating this infrastructure would create a huge economic boost for the country by allowing goods and people to move more easily. Sure enough, China continued to invest in high speed rail, and now a lot of regions that used to be hard to get to are wired into the economy, and China is seeing substantial development in these regions.
Yes, I agree. I'm not saying China is bad doing this, it's just this is what they are doing IMO.
Spending huge amounts of money to gain a foothold in AI because it helps their country is not a bad thing at all.
But my point is that the day that the USA's AI industry collapses in on itself, Chinese companies are going to also stop open sourcing their models, too.
It may sound like I'm kicking China for this, I'm not. It's just a practical take. You can't make money spending 10s or 100s of millions of dollars just to then give away everything for free and expect to make money long term. Do you agree?
I actually expect they will keep making the models open and we'll likely converge on one or two alternatives because there's really not much difference between them at the end of the day. Everybody curating their own model is a huge duplication of effort with no clear benefit. There's a reason everybody doesn't roll their own operating systems for example.
It all comes back to what I said earlier. If you treat the model as the product, then it makes sense to keep it closed. You have some secret sauce that nobody else has, and you sell it. But the reality is that nobody has a magic formula that's significantly better than what other people can figure out. You might get an advantage for a few months tops, and then other models start catching up.
And this creates involution where you just have a race to the bottom where nobody makes any money. On the other hand, if you treat models as infrastructure, and everybody contributes to the same pool of knowledge, then you amortize the cost of making a better model. The money comes from actual products that can genuinely differentiate themselves. Companies are going to seek niches they can dominate where they do a specific thing really well. That's a much more realistic path towards long term sustainability.
And that's why I expect models are going to become infrastructure akin to Linux in the long run. They're just not where profit is.
2 replies →