China really needs to speed up its construction of fab labs and start producing 14 nm semiconductor chips. The ZTE case showed Chinese dependency on American chips, which basically makes their tech companies hostage to America. I really don't understand why the Chinese government did not invest so intensively in semiconductors a couple years ago as it does now. Especially since the domestic semiconductor industry is the most important tenet of the Made in China 2025 initiative. Sure the Chinese can slap retaliatory tariffs and hurt farmers, but the US can really damage major Chinese tech companies due to this chip reliance. I think the government is finally getting the message ... better late than never, I guess. I'm starting to wonder what the Chinese tech industry impact would be if the US banned sale of American chips to Huawei in a similar fashion to ZTE ...
To accept these conditions would basically amount to unconditional surrender by the Chinese. I'm pretty sure China would rather fight to the end than agree to even one of the aforementioned conditions ...
Do you actually know what a semi-conductor fab is? and the underlying economics of semi-conductor manufacturing?
It's not Star Trek where you say make it so and things magically happen, make a proclamation and state of the art chips just materialise out of thin air!
Here's a really simplified view
- The raw material is plentiful essentially sand
- This has to be turned into cylindrical Silicon Ingots of extreme purity, its produced in environments so clean that in comparison hospital surgeries look like sewage works
- The ingots are sliced into wafers when people talk about 8in (200mm) wafers and 12in (300mm) wafers that's the diameter of the silicon ingot from which the wafer is sliced. Suffice to say the greater the diameter, the greater the difficulty in production, the reason for larger wafers is that it's more cost effective to process as the time it takes to produce the same chips on a given wafer is more less the same but the larger wafer provides more chips so greater volume is produced per unit time.
- Having made extremely pure silicon it then needs to be "doped" (essentially made impure again!) in precise a manner to achieve the desired characteristics
- A whole series of steps are then applied to the wafer, etching away layers, this is done almost like processing film (film for the millennial generation is a plastic like translucent material where light sensitive chemical reactions are used to capture scenes, further chemical reactions are used to produce the paper photographs
), further doping to change a layer's characteristics, depositing metal layers etc. The more layers a process has the more complex "chips" can be produced per unit area of wafer, so it's not just the feature size but the number and types of layers that a process offers that's important.
I'll skip the whole testing and packaging cycles
This type of manufacturing needs heavy investment it costs billions of dollars to set up a fab, it's in the ball park of the price of an aircraft carrier per fab and can take just as long to commission one! That's why the foundry business model came about as it consolidates the investment needed and optimises the capacity, i.e. you don't have hundreds of chip manufacturers spending trillions on half idle fabs. The net effect is availability of consumer electronics that are actually affordable.
Flaws can creep in along each step of the manufacturing process from the bucket of sand to the finished devices, commercially it's important that a respectable number of functioning 'chips' are produced from each wafer both from a supply and profit standpoint. Since the time in manufacturing a wafer is fixed if only 20% of the chips produced are functional then it takes a lot more wafers to fulfil an order than if 50% are thus hitting profits and also means longer time to fulfil an order which deters customers (go to vendor A get your order of 10,000 chips every month, got to vendor B it's every 2 months).
Given the challenges and costs, it's not simply a technical challenge but also a commercial one. Despite the moniker fabless a 'chip' nevertheless needs to be 'designed' (laid out) with an end process in mind, Just because a process has the same feature size one foundry's 14nm process can be very different from another's and work has to be done to switch foundries, unless of course the processes are essentially copies which would be astounding without theft!
To give a picture of the commercial realities, at the end of the day is it better for China that Huawei is able to supply and sell phones competitively using non-domestically produced ICs or limit it's production and supply less competitive phones using only domestic ICs? For one answer we can look at the Apple/Samsung relationship, Apple despite year's of law suits and competition against Samsung is still Samsung's best external customer, Samsung despite the ability to halt iPhone production by with holding supply still fulfils Apple's order and ships boat loads of components to them. Commercially the relationship is fairly win win neither side gets the whole pie but both sides gets a bigger slice of the pie than the alternatives. The obvious alternate answer is the burn it all down, I don't care mantra doing the rounds in Washington!
The economics of a wholesale US semi-conductor ban against China is no different to a wholesale Chinese ban on rare earth metals to the US, is the US spending billions re-opening mines and building smelters? well no, the last time there was the threat of a rare earth shortage every one piled in lost money and everything was shut down! So what failed for the US is obviously a winning tactic for China!
The Chinese government has already had one go at promoting the domestic semi-conductor industry with mixed results. However, in recent years the Chinese imports of integrated circuits has rocketed so the investment thresholds can be met and there are opportunities to de-risk another round of investment.
When you have responsibility for close to 1.5 billion people surely the basics of getting them housed, clothed, safe and fed are the top priorities ahead of speculatively dumping billions of dollars into semi-conductor manufacturing, in a now prosperous China the sums add up better but a decade ago, I think not. Making cases that governments should be doing X or Y and how money should be spent without the any responsibility for the outcomes is sheer hubris and dare one say a key issue with the lobbyist infested forms of "democratic" government.
We then go into the deranged fantasy world of those that make the argument "but what happens when we're at war?" there's never been a war between 2 nuclear states for good reason it might be a very short one and in this context the background radiation would mean none of the semi-conductors will work afterwards anyway. Beyond that pithy sentiment in a protracted conventional total war you wouldn't be worried about the feature size of your integrated circuits but the production of war material, logistics in supplying the frontlines and defence of the homeland you can then worry about how to rebuild your civilian industries afterwards.
We live in an era where the US, and judging by the sentiments expressed here a fair few members of this forum, thinks it can rattle a sabre and every one will fall into line, China excepted because its "special", is that actually true? I guess we'll find out!
