Even though it seemed expensive at the time, Masayoshi Son was smart to buy ARM.
* Late comers to the semiconductor industry can barely follow the trends, let alone lead. But if one cannot lead, it cannot survive in a free market. No one is going to buy a chip 3~5 years behind the latest technology when he/she has better choices.
* Compatibility issues. China cannot grow its chip with x86 instruction set, at least not the latest one, for patent issues. That further reduces the possibility of commercial success of its own chips.
* Semiconductor industry is now past its prime, the profit and salaries lower than the nascent Internet companies. Even in the US, plenty of EE students are jumping ship. In China, EE talents are doubly drained, by both American semiconductor industry, and by the local software industry.
* The Wassenaar Arrangement bans the export of many high end equipments necessary for chip making to China. While all the other countries/regions can specialize in one part of the semiconductor industry, China has to do all of them by itself, since it has no external help.
* The world politics dictate that Chinese firms are never going to buy ARM. They have the cash and the willingness, but western countries simply won't approve such buyout.
Forcing your enemy to invest heavily in such key sectors is pretty stupid. Eventually, one way or another, Chinese will master the state-of-the-art. A few good examples here - LCD industry, weapons.
It also worth pointing out that the existence of the Wassenaar Arrangement relies solely on the fact that the US is the biggest economic and military power in the world. Not sure how long the US can continue to maintain it.
Given other more developed countries have a technology advantage that they're not sharing and they do not face a demographic crunch, its possible China will find themselves perpetually behind in many areas.
No, they don't need to to begin with. They are smart enough to not to parrot late USSR.
Instead, it is seems more evident with each day that they use counterstrategy - sealing West's advancements in what will be the stem of state-of-the-art tech tomorrow, for cheap, instead of fighting for already substantial, but heading for decline things-that-matter markets.
That used to be a working strategy. But now that ZTE was sanctioned by the US and almost crumbled overnight, it is obvious that the United States is capable and willing to throttle China on these "already substantial, but heading for decline things-that-matter markets". So China has no choice but to master the state-of-the-art of every important sector.
that is not very smart. the ZTE drama is a wake up call for Chinese, the message is loud and clear - US products are not reliable in term of its supply, it can be cut at any time for some stupid political reasons ordered by a moronic US president like Trump.
if you look at the whole story, it is push the lower limit of the business IQ - - forcing your competitor to invest in high profit sectors - telling your customers to rely on less American products by convincing them that the supply won't be reliable
US deserves a better president.
This would have been pretty uncontroversial under any other President.
Read this for more detail on ZTE's malfeasance: https://www.justice.gov/opa/pr/zte-corporation-agrees-plead-...
Generally this makes sense, but not in China’s case where they require all foreign ventures to team with a Chinese company who then steals the foreign partner’s IP. No way a Western semiconductor company could set up a fab in China without creating a competitor 5-10yrs later using their own IP against them.
Thus the Wassenaar Arrangement makes sense, it prevents any Western semiconductor company from gaining a short term advantage over its competitors by leveraging Chinese cost efficiencies, incentivizing all of them to self-destructively follow suit.
China is also capable of figuring things out from first principles, or espionage from non-partners. An example of this was the decision by the US to exclude China from the ISS - they built a couple on their own (no partners) and will be there after the ISS de-orbits or privatizes.
Not if your enemy has a habit of reverse engineering and copying everything they buy.
I recall that China was has a competition for foreign companies to sell them attack helicopters. They had trials and evaluations, and settled on a South African one. You'd assume they made a big order, right? Wrong. They placed an order for a single copy, which the South African company wisely declined.
I can't seem to find a source for the above story, but I think I read it in Wired maybe 5-10 years ago.
kinda of true, assuming you want to be more than a fabless company. But spinning up a fab 1 gen behind can be profitable, especially if it means higher yields (and thus lower costs).
true
* Still lots of money to be made at the forefront, i.e new chemistry, new lithography, better process control, etc. Lots of money in layout and software. Pretty decent money in RF work.
China has proved itself more than capable of replicating anything you put out there -- all it takes is some plans and a few key engineers, plenty of whom are Chinese nationals.
I agree.
Past its prime in what sense? We'll need them for the foreseeable future. Salaries and all can be fixed virtually overnight, with funding...like this $47 Billion.
https://en.wikipedia.org/wiki/Loongson#Loongson_3
Out of curiosity I tried to buy the Loongson 3 one year ago on Taobao (Chinese ebay/Amazon). It was basically impossible to buy except for underpowered old versions.
As far as I know the company has been called a scam by some Chinese (just sucking gov incentives etc.)
If you are interested, not very powerful MIPS boards are available: https://creatordev.io/ci20.html
I guess the company "imagination" is for sale too. There were some issues last year.
Similarly, in the US you can make something that was patented in China as long as it isn't also patented in the US.
They did. Beijing handed out massive subsidies, companies started building fabs, and now those fabs are starting to come online in 2018[0]. Fabs take time to build.
[0] https://epsnews.com/2018/04/26/three-chinese-companies-to-st...
CSMC and Huali for example are a ghost towns (i worked few minutes away from both)
They continued to throw tenths of billions at the problem, till they managed to make a minimally well running fabs for few niche markets.
Like already someone commend. They did! It's called "Big Fund" and started in 2005. This is just another round. They already have been multiple rounds as far I know.
These days you are require to prove. And there are people qualify to judge whether you are actually getting somewhere.
That said, Semi Conductor is expensive, and requires lots talents and a whole IP ecosystem around it none of which China has. They are trying to paid to get these component and talent, copying and learning from it as we speak. But whether that works and how long it will take we have no idea. After all Semi moves at a very fast pace, by the time they caught up with 14nm, the world would have moved to 5nm.