The CHIPS Act won't end US reliance on foreign foundries With all the fanfare and foundry expansions around the signing of the $280 billion CHIPS and Science Act, you'd be forgiven for thinking that the semiconductor shortage will soon be over, and the US will emerge as a silicon powerhouse rivaling that of Taiwan's TSMC and South Korea's Samsung and SK Hynix. But that is a …
While they were nearly that enthusiastic in the recent past, running fabs out in chandler taught them how to recycle fab water. New water intake is now a fraction of the old open loop operations that used to kill fish in east Fishkill.
As it turns out, after fighting it for years, it's cheaper to re-purify water coming out of many of the water intensive stages than to de-ionize and buffer river or well water in the first place. So don't go applauding their green initiatives too much, they fought it for probably 5 years past the point it was cheaper to implement, though that was probably more about aligning the upgrades with other large process improvements than and bond villain motives.
Never mind the water, the Texas "power grid", and we can use that term loosely, won't hold up to the growth they are trying to impress upon their urban and industrial areas.
Texas is becoming a perfect example of putting money first, just about everything else secondary. But these are the voter's choices, so let them eat cake.
the Texas "power grid" [snip] won't hold up to the growth they are trying to impress upon their urban and industrial areas.
How about a brand new power plant to go with it? I suggest nuclear, followed by oil, natural gas, and coal.
(It is common practice for planners to REQUIRE investment in infrastructure for new construction)
"I suggest nuclear, followed by oil, natural gas, and coal."
Petroleum oil isn't used for large scale power production in the continental US. It's too expensive. Some refineries use bunker oil to generate power to run the refinery since it's the least valuable fraction.
Nuclear is a good idea and especially modern designs rather than the really old fashioned PWR models that have been built for last 50+ years. Natural gas is good for peaking and I think that it's good that coal is on the way out. Wind and solar are good for uses that can adapt to an intermittent supply. They can be very good for charging electric cars owned by people that use them to commute to a regular job and can travel all week or more on a full charge.
Still, it takes many years to build an oil, natural gas or coal power plant, and DECADES to build a nuclear plant. Never mind the environmental impact, the support infrastructure, the interconnects and everything else that is necessary to be created just to get the power plant actually placing its power on the grid.
IMHO, and you may disagree, but in the long-term Texas is headed for a fall. Their thick-headedness about, well, anything that doesn't fit their predefined agenda of how their world [should] work will end up causing massive infrastructure issues in the near future, considering their desired growth. People are running to Texas from California under the belief that the free-market capitalism that Texas supports will be their saviour, but it is worth remembering that Enron was Texan, too (foreshadowing...)
That should be "the free-market capitalism that Texas claims to support". They'll happily send men with guns to stop Safeway from selling whiskey on Sunday morning; and forget about opening a card room that competes with the state-run numbers racket. Even the gun laws aren't very permissive compared to hard-core red states like (checks notes) Washington.
"They'll happily send men with guns to stop Safeway from selling whiskey on Sunday morning"
Which makes sense in their Christian Nationalist worldview. :rolleyes:
Ever read the web comic Poisoned Minds? A UK comic by Alan Foreman, and he actually forecast the downfall / breakup of the U.S. over a decade ago. With Texas turning into a fundamentalist regime.
Imagine that??
Then there's always Arizona, the Phoenix area. Got a lot of investment there and its growing. The only problem with Phoenix is that its not possible to live there for much of the year without extensive life support systems (A/C is only part of the story) and that whole area's out of water.
Intel / TSMC /etc are going to take the subsidies to build next-gen fabs - the ones that make the most profit.
Nobody is going to be building the cheap, high volume, low profit parts that automakers need. And even if you did, the dicing, packaging and assembly of these $0.05 parts isn't going to happen in Austin
>Today's cutting edge will be tomorrow's old process node.
That is a major concern.
The EUV magic to get 3-5nm is highly involved with the chip design. It isn't like just taking 20-40nm parts and getting a 50-100x density for free.
It might not ever be cost effective to make cheap older parts on these systems even if the fab cost was already paid
"Today's cutting edge will be tomorrow's old process node."
To some extent that's true, but it's getting to the point where further improvements in resolution aren't going to be all that spectacular. It's the same as hitting a wall on clock speeds due to the speed of light and the distance from one side of the chip to the other and from the CPU to memory, interfaces, graphics cards, etc.
take it one step further .. why do we need those old tech parts ?
because the auto mfg's want to keep an old design and not go through the process of redesigning their products. this is a self inflicted wound. they CAN do it but won't . That is what you get when decisions are made behind an accountant's desk and not an engineer's desk.
I have 0 pity for the car mfg's at all. Adapt and evolve or die. Their choice.
I was thinking something similar, that we need fabs for resistors, capacitors, SMT inductors, discrete semiconductors, and small scale integration. I somewhat recently ran into supply issues with a small scale integrated circuit where there was NO pin for pin substitute. I managed to find one with inverted logic and fixed it in the firmware.
Still prior to 2020 all of those parts had good inventories at distributors and reasonable lead times,. After China started their sudden and massive lockdowns, it was like a crap shoot as to which components would run out any time soon. these boards are built in USA and Mexico so it would be a GREAT advantage to have fabs in N. America at least.
Not some Fortress America nonsense. Funny how the plan of last resort when fighting the Fascists in WW2 became the favorite talking point of Neo-fascists as we close in to 100 years later.
Big hint, the people who push hardest for "Fill in the blank" independence are Isolationist Nationalists. They know that it is best flavor of Kool-aid to cover up the bad taste of the recession and decline it will always cause. Then they sell the glorious leader as the only person to protect the great nation from the enemies that destroyed the economy and threaten it's borders. By the end they switch over to the grape Flavor aid.
So let's not loose to much sleep over the fact that our foreign allies and trade partners will still have a functioning economy to look forward to.
..and for those who wonder why we are playing nice with them, remember, they will be the front lines of the brutal proxy war that an open conflict with China will result in. If it's not attractive to stay on our side after a major conflict for fear of getting shut out of the future tech economy, Korea and the pacific rim will have to think about what side they want to be on.
The better position on the game board now is worth a few billion up front, even if you ignore the significant returns this investment will probably net us over the last 20 years. It'll be trillions later.
"And the obvious national security concerns involved. When your military requires so many semiconductors, not being able to make them is a problem. Same with producing steel, oil, food, etc."
True, but when some government agency decides that a certain scale process has "national security" implications that require companies to apply for permission to export their semiconductors, it can stop making sense to make them in country. Compounding the problem is those government wonks have no clue what's state of the art today and what was state of the art 10 years ago. Regardless of how much money the US will give away, it might still not make sense for companies to build cutting edge fabs on American soil only to have eight or nice agencies telling them what to do, how to do it and who they can do it to.
murkins have never fort a war on there own turf, always raping someone else, and the propergander machine always makes them smell so sweet,
come on in suckers , we'er going to make A murker great again,, at someone elses cost!!
Please check your history. A simple Google will do. More than seventeen wars have been fought on U.S. soil (I gave up counting, but hadn't yet got to the War Between the States).
That said, it does seem that most of the modern wars the U.S. has been in have been money-related.
To be more comparable to the density of TSMC's. Since terms like "7nm" or "3nm" no longerhave any meaning as far as actual physical dimensions of the transistors, it is probably fair to do so.
Thus Intel's latest "10nm" is called Intel 7 and is comparable to TSMC's N7, and their upcoming process is called Intel 4 and is comparable to TSMC's N4 process which has recently entered mass production. They claim (but it remains to be seen) their Intel 3 process will enter mass production before the end of next year, which would be comparable to TSMC's N3 which enters mass production around the end of this year.
So while still behind, they aren't quite as far behind as the article makes it sound.
The TFA is wrong where says, "But Intel, as massive and influential as it is, doesn't have a lot of experience producing the kinds of microcontrollers needed for things like engine control units, air conditioners, or even your coffee maker."
Intel has been making microcontrollers for decades! There are 21 microcontrollers in the MCS-48 family, 17 in the MCS-51 family, 5 in the MCS-151 family, and 20 in the MCS-96 family. Most of the 8061s they make are sold to Ford.
I took that as "Intel doesn't have a lot of experience of manufacturing currently used microcontrollers", especially when compared with Microchip and others who specialise in that field.
CHIPS could be described as "building rocket-boots for the elite while everyone else is desperately in need of cheap sneakers".
So US companies are doing what they've always done: abandon the low end, low margin work to the overseas companies. Possibly the result of management liking shiny sexy stuff instead of what people really want? The only industry that can't flee the country is food processing, where for some strange reason, low margins coupled with high volume work just fine.
Came across this on youTube last night....
https://youtu.be/5Ge2RcvDlgw
It gives some of the science and background to the EUV 13.5 nM light source that's needed for fine grain semiconductor lithography. This is just one part of the whole, it touches on the rest of the machinery, which in turn is only one part of the entire process needed to economically make high density semiconductors. It is a strategic product but its also a global one --- there's a lot more involved than just a machine here, a facility there.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
