So my guess would be in the next decade, the real money will be on the design and IP not the manufacturing. ( excluding the bleeding edge half-a-nm stuff )
Intel has fallen behind because it failed to deliver its 10nm (now called 7nm) process in 2015. Intel tried-and-tried, but failed, to the point where supercomputers (ie: Aurora) were delayed because of the manufacturing mishap.
Just because everyone is __planning__ to open a fab doesn't mean that they'll be successful at it. If Intel fails to reach 5nm (TSMC-name), then TSMC will win.
In fact, TSMC has already started making 5nm chips called the Apple M1 Pro.
---------
Intel has finally delivered 10nm, but they were 5 years late. If it takes another 5 years to deliver their version of 5nm, then TSMC will be at 3nm by the time Intel reaches 5nm
-----------
Its all about manufacturing actually. All of these chip designs can be ported over to different manufacturers, so long as the different manufacturers are __CAPABLE__ of implementing the designs.
Everyone is buying the same machines from ASML. Machines aren't the issue, its process. Its nitty gritty manufacturing details that no one will talk about (because those details are the "moat", the secret sauce that makes TSMC do something that Intel can't).
It looks like Intel has figured out 7nm, just much later than expected.
TSMC has competent leaders like Morris Chang and Mark Liu.
Intel had lost it's way after Andy Grove gave up the helm. Pat Gelsinger grew up under Gordon and Grove and the old guard is back.
Intel is institutionally paranoid, and Grove is part of that fault. The old guard, even in engineering, is retiring/dying and have not groomed successors because that could have threatened their political position.
Can Gelsinger really fix this? I suspect probably not.
Absent any geopolitical difficulties... but that's a whole other discussion.
It takes a lot of water to make the ultra pure water needed to wash the wafers. Without access to lots of water they will too will have issues.
PRC will have enough military strength to attack Taiwan and challenge US Carriers. We don't know if they'd win but... enough strength that they're a real threat... at least near China (with assistance from land-based hypersonic missiles and whatnot). PRC probably doesn't have the military strength to attack Arizona.
China will have a Navy that (by tonnage) rivals the US's Navy in a few short years. US still will have more capital ships but there's big, unanswered questions about what the proper tactics will be in the next great naval battles.
Would we rather have the US's "Supercarrier" strategy? Big ships with lots of money put into each one? Or would China's "little ships" strategy be superior?
Really: one question. Will China's hypersonic missiles truly be a capital-ship killer? If their weapon can truly kill a US Supercarrier, (or more specifically: if China __believes__ it can kill the US Supercarriers) then China will probably attack Taiwan.
China is building a navy for _some_ reason. Taiwan seems like the obvious target.
------
China's increasingly large Navy is a big enough deal that the USA has given away its nuclear submarine secrets to Australia to counter-act the threat. That doesn't really change the picture too much (US always had nuclear-submarines available), but its clearly a degree of saber-rattling going on right now between US and China.
we had an actual intel plant manager who actually ran one of intel's fabs on one of the previous threads about this and HNers of course deigned to tell him his business, as usual. Love it.
Geopolitics have been figured out. TSMC's next 5nm plant will be in Arizona. https://appleinsider.com/articles/21/06/01/tsmc-starts-const...
The current shortage is making countries like the US realise that semiconductors are a critical commodity for the function of their advanced economies.
The US wants fabs back on their soil and will probably be willing to burn money to make it so.
Will the US be satisfied with just having foreign operated fabs on US soil? Or will the US want home growth (and firmly under US control) companies like Intel to have similar capabilities.
Frankly, IMHO things in the long term aren’t looking that great for TSMC. Some of their business is going to get eaten by “local” fabs - even if they can’t do leading edge nodes they will eating into TSMC’s business on the lower end.
There is also the danger of overcapacity destroying any profitability in the fab business altogether - guess who will get bailed out by the US government if that happens; hint: not TSMC.
The boom/bust cycle is nothing new in semiconductors. It's a fairly regular thing. Also, keep in mind that consolidating and repositioning during downturns is how TSMC became TSMC.
In general, if semiconductors become a national security issue, I think there will be a fundamental change in the nature of the business. It will not be business as usual for TSMC.
Maybe, who knows what the future will bring.
But I've heard this before [1] and once recessions hit, and budgets get cut, that kind of thing quickly becomes less tenable.
And then someone realizes that it takes a constant stream of incoming parts to keep the equipment inside any factory running for any meaningful amount of time, and that supply is not protected, and neither are their suppliers...
But I agree, things will not stay as they are. The only thing usual about this business is change.
[1] The US semiconductor manufacturing sector was under serious assault by the Japanese in the 80s. That was my first "trade war" in this industry...
https://www.washingtonpost.com/archive/business/1985/09/22/s...
Semiconductors could end up like the US defense industry, which the US feeds with a continuous amount of money (recession or not) and is not outsourced for obvious reasons - but its products are sold to others; other countries outsource to the US. The US buys military tech on occasion but rarely ever directly buy the hardware in significant numbers - they will manufacture it in-house. It's great if it turns a profit but that's not the point.
Every country has their own defense industry if they can afford it.
That said, it remains to be seen how seriously the US government is about this. The $50 billion USD bill hasn’t passed through Congress yet. But if the US government does decide to make huge moves, things might change in a big way.
TSMC is no doubt pretty worried hence their setting up of fabs in the US now in an attempt to hopefully ward off US government action.
TSMC's Q3 shown bulk of their revenue come from higher end node. 7nm along is at 34%. the only other options for high end node is Samsung and Intel.
I don't see any long term problem for TSMC if they keep going at its current pace.
Not many foundry can do high end. GlobalFoundries vow out at 7nm a long time ago.
https://investor.tsmc.com/english/encrypt/files/encrypt_file...
If anything, I'm far more worried about the PRC exercising the soft power it's slowly been building in eg. part of Africa.
China is building new ships today: roughly 3 aircraft carriers, and a huge number of Destroyers that will (on a tonnage basis), outnumber the entirety of the US Navy. They aren't capital ships, but a mass of lower-quality ships would probably still work if those hypersonic missiles can really kill our capital ships. (A big question in of itself: do China's hypersonic "ship killing" missiles work? If so, can the USA's defense systems shoot them down? If not, then why wouldn't China attack Taiwan?)
Meanwhile, the US Navy is just about to upgrade over the next 15 years. This means that a number of our ships will be docked, and the ships that are remaining will be older designs.
If China times it correctly, it can attack Taiwan and force the USA to defend with its obsolete ships, and before the new ships come online. In effect: the USA has already "spent" our money / resources on the upgrade process, but we haven't gotten the benefits of it yet.
----------
This is a once-in-a-generation opportunity to attack. Once the USA's fleet is upgraded, the opportunity closes (as USA will have newer / better technology ships, while China's old Carriers/Destroyers will be at the disadvantage. And China won't reclaim the advantage until it spends a huge amount of time and money on their own upgrade cycle).
Furthermore: this "delay in upgrade cycles" was caused by Bill Clinton's budget in 1996. It takes many years for a budgetary issue to come up, and the USA has spent huge amounts of money on its military from 2001 onwards (because of Afghanistan / Iraq). As such, no such "upgrade gap" will exist for 20+ years at the minimum after this opportunity passes.
Will China be willing to wait another 50 years for another opportunity like this? Or will China actually go for it? And remember, China has already dedicated incredible amounts of money towards building a new fleet. China's fleet is a "sunk cost", their only decision now is whether or not to use it.
It turns out that the human ability to make strategic blunders is very possible. So if your best argument is "it'd be a strategic mistake to do X". Well... history is filled with horrible strategic blunders.
It'd probably be against American interests and a strategic blunder to divert forces from Afghanistan into Iraq on an notion of WMDs for example. Everyone makes strategic blunders all the time. I don't fault people for making them, I just try and study them and understand why it happens.
----------
There's something fascinating about war machines, like carriers, hypersonic cruise missiles, and destroyers. You have to see them in action to understand the intoxicating effects of these machines.
Once you understand how these machines can hypnotize a person... how they can make you feel like you can accomplish anything... then, and ONLY then, will you understand why Wars get started.
China's got Hypersonic cruise missiles. They know it. We know it. China knows the USA doesn't have them (!!!). That knowledge is intoxicating. China sees that as a military advantage. Will they be intoxicated by that knowledge to the point where they attack Taiwan?
Well, no one knows for certain. But they've looked at those missiles and have __already decided__ to build more ships than the entirety of the US Navy. So what does that tell you?
This is an important point and why I believe Wolf warrior diplomacy runs contrary to China's national interests. It has caused countries who previously viewed China as merely a competitor to view it as a direct threat and to take appropriate action. You know you've gone too far when your actions have resulted in Korea and Japan aligning with each other about the need for closer ties and to align with the U.S. on regional security.
The US officially doesn't rule out escalating from conventional to nuclear war against other nuclear-armed countries. Even if the possibility is remote, I imagine that nobody is suicidal enough to want to test the possibility.
There's a big difference between nuclear powers lobbing 100kilo-ton bombs at each other's navies in the Pacific, and nuclear powers using 10-Megaton bombs at each other's cities.
Given that China has decided to make a large mass of cheap Destroyers, what do you think China is thinking about this next war? Tactical nuclear weaponry would favor the one who has a more dispersed naval force, and be a bigger problem for USA's multi-billion $$ supercarriers, would it not?
-------
Lets say China nukes a military Supercarrier. Will the USA respond by nuking Beijing? Of course not. Knowing that USA is tactically in the disadvantage from using nukes (due to the composition of our fleet vs the Chinese fleet), we know that USA won't escalate to using even tactical nukes.
Also, accidents happen, and it's not clear how secure strategic weapons are against everything from a zealous airman to a submarine cut off from its command and having to make a judgment call. Daniel Ellsberg's book (The Doomsday Machine) suggests that this kind of accident during a crisis has come close to triggering a full nuclear war several times.
To summarise, I'm sure the tactics look good for China, but escalating to nukes still seems like a suicidal move, considering the risk of accident or miscalculation.
But I can conceive a hypothetical Chinese Admiral who disagrees with the both of us. And that's all that it will take.
https://en.wikipedia.org/wiki/DF-21#DF-21D_(CSS-5_Mod-4)_Ant...
The DF-21, available maybe 10 years ago, would have the range to hit ships all the way out to Japan or so. That's really all you need to know: China is trying to develop (or already has developed) "ship-killing" hypersonic missiles.
But of course, no one knows if they work vs US ships (can US Ships deflect the missile? Do we have flares or something that the missile will mistake for a ship? Can the US shoot down the missile before it reaches a ship? No one knows).
-------
For China's military composition: https://media.defense.gov/2020/Sep/01/2002488689/-1/-1/1/202...
You can see in the Eastern/Southern theaters of China, they have roughly 500,000 Army troops vs Taiwan's 80,000 troops. (China has a military of over 1-million, but most are deployed elsewhere). China also has:
"The PRC has the largest navy in the world, with an overall battle force of approximately 350 ships and submarines including over 130 major surface combatants. In comparison, the U.S. Navy’s battle force is approximately 293 ships as of early 2020. China is the top ship-producing nation in the world by tonnage and is increasing its shipbuilding capacity and capability for all naval classes."
So PRC already outnumbers the USA in terms of ships. In a few more years, PRC will outnumber the USA in terms of tons. China is definitely going for "build more" and not necessarily "build better". There hasn't been a modern naval battle since like WW2 between major powers, so no one really knows if this is a good or bad strategy. But its certainly a strategy.
USA of course has our 10x Nimitz supercarriers + 1x Ford supercarrier (going through our carrier-upgrade cycle currently, +F35 project I'm sure you're aware of). USA has a number of "Amphibious assault ship", which are sorta-kinda like carriers (except for the Marines instead of Navy), but do contain VTOL aircraft like F35-B variant.
I don't think anyone really knows for sure how the battle will play out. The US ships (ie: the "Supercarriers") are incredible war machines, and even the smaller Amphibious assault ships would probably go toe-to-toe with even the Chinese carriers. But how does that play out with China's missiles and airforce?
How many rabid PRC-haters do you think Taiwan harbours? Well, it's more than one...
von Braun worked for the Nazis to make missiles during WW2. Then was picked up by USA / NASA to make missiles / ICBMs / Space Program.
It turns out that a surprising number of engineers are rather flexible when it comes to ethics. They just want to work on cool things.
...after a couple of years of unrest and social instability. Which as simonh pointed out (https://news.ycombinator.com/item?id=28959591) is exactly what the CCP does not want.
Knowing that the PRC tends to take the long view, I don't think they'd take the chance regarding Taiwan.
If China wins Taiwan vs the USA, there wouldn't be any unrest or instability. The Chinese people would be proud of their accomplishment, and the rise of China as a military power.
Like... what "instability" occurred from Russia's annexation of Crimea? None what so ever.
They do, and that's the problem. The shot clock is ticking down quickly: fertility rate in PRC just hit 1.3 births per woman. Total population is projected to halve by the end of the century. The current cohort of military-age males is shrinking, and there won't be another one to replace it. The PLA will never be stronger than it is today.
If Xi wants to take Taiwan, he has to do it now, because he's not going to get another chance. PRC manufacturing output is cratering due to power shortages and the global shipping and semiconductor convulsion, so the economy is going to get clobbered anyway. They're already doing recon flights through the ADIZ and running freedom of navigation naval patrols. If there's going to be a Taiwan annexation, it'll happen in the next 12 months.
The US Navy is hapless and the Taiwanese army is supine. Evergrande is collapsing and there's already a wide array of tariffs and sanctions against PRC leadership. They accidentally caused a global pandemic and shrugged off any political ramifications. They've already taken half a dozen hits this year and survived. Why not take another?
The Taiwanese military has had decades to prepare for a Chinese invasion; it would be surprising if they didn't have a similar setup in place.
The CCP decided many decades ago that economic growth was absolutely necessary for them to stay in power because it is required in order to maintain social stability. An economy with hundreds of millions of unemployed peasants in the countryside, and feeble industries was not sustainable. This is why they purged a bunch of ideological marxists and embraced communism with Chinese characteristics, AKA Capitalism.
For a long time the CCP maintained that 7%+ economic growth was necessary in order to absorb the unused labour pool and maintain social stability.
Needless to say a war with Taiwan and the US, plus blockades by the rest of the world would be a somewhat risky situation from a social stability point of view. It would be a complete throw of the dice, and for all Chinese people love gambling with money, that's not the CCP way with politics.
We know the USA would respond militarily. But how many other countries would be pulled into the conflict? Germany still trades with Russia despite the Crimea annexation, right? Clearly USA / UK / Australia are making plans but is there anyone else who'd join in the fight?
Like, realistically, what would happen under a Taiwan scenario? Do you think McDonalds and Apple will just leave China? Will Tencent games be banned in the USA? Will USA really make League of Legends and Genshin Impact illegal during the war?
Or, would the world continue as usual? Would US Video Gamers continue playing their games, McDonalds continue China operations / Apple continue making their iPhones while the whole military stuff happens? Frankly, I see this as the more realistic scenario.
We can already see how video gamers reacted to Hong Kong: with a bunch of companies capitulating to Chinese demands.
I'm in no way justifying what the CCP did to Hong Kong, but frankly it was inevitable. That is not the case with Taiwan, which has a population and democratically legitimate government completely committed to independence.
An invasion of Taiwan would be much more similar historically speaking to the German invasion of Poland, or the Korean war. It's utterly and completely unacceptable, and would be legitimate grounds for total war. Should China succeed in conquering the island it would not be possible to counter-invade, but an indefinite blockade of transport into and out of China and total interdiction of it's air and sea traffic both civilian and military would be entirely proportionate and the minimum I would expect. See the blockade of North Korea for an example. We'd buy our phones and video games elsewhere.
And China outnumbers the US warships ship-for-ship currently, and China is outbuilding more warships ton-for-ton than even the USA.
China also has a larger (though inferior) air force, as well as a huge number of (untested) missiles. But the USA won't have many bases to stage our Air Force from (Taiwan is so close to China, and so far away from Japan / South Korea). So really, its China's Airforce + China's Missile-force + China's Navy vs the US Navy + US Marines + maaaayyyyybe a very stretched thin US Air Force.
If the USA wants to enforce a naval blockade, they'd have to win a war against China. That's a lot of advantages to China frankly.
I'm certain the US could beat China in a "fair fight" for Antarctica. But the issue with the Taiwan scenario is that the Chinese military is growing ever stronger, and its conceivable that China could beat the USA in that region. If China does happen to win militarily, then a US-sponsored blockade would be impossible.
There is simply no way China is going to contend with the US globally. Zero trade with the US, Europe, Japan, South Korea and Australia is a given. A sea blockade of trade with Africa, South and Central America, and Australia is fine. That leaves land and air trade only with Asia. How are they going to get oil? China's not going to establish military domination over the Middle East any time soon.
As a Brit I don't see us staying neutral, or Australia. Hence the nuclear subs Australia is buying. In a conflict like this, they would use them.
There may be a short term benefit but if they get cut off from ASML (Europe), Applied Materials (US), etc. it's going to be tough to keep their machines in service and stay leading edge. And those companies have very deep international supply chains too and may also get sanctioned.
You sound like you have a dog in this fight.
TSMC have a lot of fabs in Taiwan. that doesn't mean TSMC as a business can't build fab elsewhere.
TSMC is already building a fab in AZ USA and reach an agreement with Sony to build a fab in Japan. TSMC also have fabs in China and India govrt is looking to Taiwan for a fab in India.
This is not a recent development. TSMC has been mass producing 5nm chips for slightly over a year now. The first 5nm chip on the market was the Apple A14 (iPhone 12 line) released almost exactly one year ago, and soon thereafter (November 10th, 2020) the first M1 Macs.
Samsung has had their ""5nm"" chips out since early 2021 with the Exynos 2100 and Snapdragon 888. It's in the ballpark of TSMC's process but not as dense.
Intel hasn't even announced their first mass market 7nm (or ""Intel 4"", sigh) products yet. ""Intel 4"" is inferior to Samsung's and TSMC's 5nm processes on paper, and it'll likely not even be next year that we're going to see products with them.
I don't think Intel's foundries will be competing on the leading edge of the market. They'll have be trying to have value focused production with their 14nm and 10nm processes, I would guess. There should be plenty of demand there still, but it'll be a good while - if ever - before Intel is leading again.
According to transistor density information this is false (https://en.wikipedia.org/wiki/5_nm_process)
TSMC 5nm density: 173 MTr/mm^2
Samsung 5nm density: 127 MTr/mm^2
Intel 4 (estimated): ~200 MTr/mm^2
Got a source? I'm willing to change my mind, but the data doesn't agree with your assertion from my point of view.
Samsung's process has always lagged TSMC and INTC. It's part of why the 3080 is so hot/power hungry.
What do they win? More maket share, more revenue, more profits, better margins, competitors goes out of business?
I know very little about this industry, but is the market big enough for all players to 'win'?
https://www.reuters.com/business/intel-details-mixed-source-...
> Intel said Thursday its "Alchemist" graphics chips will be made by TSMC using the latter's newly named "N6" chipmaking technology, an upgraded version of its "N7" technology.
I mean, I guess they're already "winning". Intel's own chip-making group has begun to use TSMC. Hopefully Intel can figure something out about what its like to be a customer, so maybe they can be a better supplier?
But yes: customers, market share, and knowledge. By having more customers, TSMC can perform more experiments on more leading edge nodes, like 3nm or smaller.