A pedantic correction- ITAR (International Traffic in Arms Regulations) is a Department of State control and is solely focused on defense and military related technologies. It does not consider dual use technologies. The Department of Commerce controls the EAR (Export Administration Regulations) which are concerned with dual use and commercial exports not controlled under ITAR. ITAR’s USML (United States Munitions List) is pretty easy to interpret as far as what is controlled. The EAR’s CCL (Commerce Control List) is a morass of specific technical metrics, often out of date and sometimes contradictory. If you are a commercial company, the EAR is overwhelmingly the export control you will need to use.
Moving on, it looks to me like the USG is trying to duplicate their prior ITAR success in commercial satcom in the chips, AI, and clouds. In the early 90’s, all satellite technology control was moved int the ITAR regulations. At the time, the U.S. controlled 80% of the global commercial satcom market. By the early 2000’s, the U.S. market share was 20% and “ITAR free” satellites and satellite components were fully competitive with the best the U.S. commercial satellite industry could produce. While the regulations have been made more sensible in the past 10 years or so, the damage has already been done.
A few things in the past couple of years have illustrated just how broken the U.S. military/industrial complex has become:
The shocking discovery that manufacturing of ammunition has been allowed to degrade so much that it will take longer than the duration of WWII to replenish stockpiles of some missiles and artillery. The west’s enemies are noticing.
NASA’s reliance on defense contractors for SLS has highlighted just how inefficient and bloated they have become. The same for the Commercial Crew program.
Boeing can’t do anything right. They’ve lost money on every fixed contract they’ve bidded on, their products are riddled with problems and kill people. The MAX 9 door flywpaway last week was just the latest in numerous identified manufacturing errors, testing shortcuts, etc.
The Littoral Combat Ship program was a disaster. These were the ships that were supposed to fight in coastal waters. The program has now been cancelled, as these ships can’t survive the modern coastal battle space. And they are brand new as military ships go.
In general, the defense contractor industry is a mess, everything is probably at least twice as expensive as it should be, development times are way too long, and private industry is starting to leapfrog them.
Time for the military to move away from the behemoth old military suppliers and start farming out development to newer, hungrier companies not burdened with layers of useless executives and bean counters.
Of course not. Where would all of those Majors/Commanders and Colonels/Captains go after retirement if there aren’t a bunch of defense contractors out there?
The DoD would desperately like to tap into commercial companies. The only thing preventing this is…the DoD.
Under the current acquisition regulations, anyone doing business with the DoD is going to have to comply with the DFARS (Defense Federal Acquisition Regulation Supplement) regulations and meet specific Mil Spec requirements. The paperwork, audit responses, legal waivers (did you know there is a requirement that all materials used for DoD procurements must be "American made and sourced. That means if you sell a steel hammer you have to certify that the steel is all American made or get a waiver. How much are you going to charge for that hammer if you are the vendor?). DoD contractors are the only ones that have the infrastructure to deal with Federal procurement requirements and are willing to live with the low profit margins on those procurements.
The DoD would love to buy from any commercial vendor that will provide their products at commercial prices but meet the FARS and DFARS (as well as certifying Mil Spec). Oh, and pay out of their own pocket for all of the security infrastructure required.
Acquisition reform is a constant topic, but no one is willing to actually do anything more than apply cosmetic changes. Until that happens, the defense contractors will fill a very needed part of the market.
A historical example:
The ENIAC computer was working in December 1945, and delivered to the US Army Ordnance Corps in July 1946.
The commercial version of this computer, the UNIVAC I was delivered to the Census Bureau in April, 1951 (4-3/4 years later). The first deliveries to several non-government, commercial customers were in 1954.
So in the early computer field, there was a lag of about 8 years from a working military version to a working commercial version delivered to private industry. (Much of this was due to the high cost of the machine, and the need to cost-justify this investment by increased profits.)
None of these things show anything new or how “broken” the US military/industrial complex has become. One of your examples is plain wrong, two don’t even involve the military, and I could provide you with dozens upon dozens of examples of failures and disasters on the part of the US military far worse than the LCS. Just google “MBT-70” or “MGM-51Shillelagh” for some expensive failures that either were canceled or worse, were put into production and relied upon despite their being failures as weapon systems. Or if you want to see something in action and how it affected an actual war, have a look at the Mark 14 torpedo in WWII, which had it not been such a failure could have shortened length of the war against Japan by a year or two. Drachinifel on the topic:
This has been done to death and back again multiple times. There is nothing shocking about it, and the West’s enemies aren’t noticing it, they’re experiencing the exact same problems. There has been a shell crisis in every major war since indirect fire artillery became a thing, and it usually becomes a serious issue about a year into the conflict. For examples:
You must be unaware that China has plenty of fabs today. The important step by the US is not banning the purchase of chips, but keeping China from getting access to the most advanced lithography technology. That Huawei can make a 5 nm chip is irrelevant since it would have been designed long before the current bans, and in any case 5 nm was not even leading edge when I retired 7 1/2 years ago.
TSMC is way ahead of this, and in any case the big advances are no longer just shrinking feature sizes, but in innovative transistor designs.
Back when Sandia was run by Bell Labs I worked with them. They had licensed Intel processor designs and had a fab on-site specifically to build rad-hardened reliable chips with technologies which had been pretty much abandoned by the civilian market. My colleagues always said it was for satellites and space probes, but we know what the chips were really for.
I spoke at a couple of conferences for electronic testing in the military, and was on a committee where some of us tried to get some of the more recent technology into the military, for improved diagnosability. It was interesting. Not just jets but even armored vehicles had technology that recorded everything, so that when it failed it would be easier to diagnose. And I mean when. One radar system someone talked about during a panel I was on had a mean time to failure of like 12 hours. And that was considered an improvement.
They also bought lots of COTS (civilian or common off the shelf) hardware because it was more advanced and cheaper than things specifically designed for them. Obviously only in places where it was feasible. I owned the processor reliability numbers for my company at one time, and we were seeing billions of hours between failures.
BTW I’m working on a book about how military innovations are transferred to the civilian world, and the furthest ahead of anything I can think of was GPS, but that was because of deliberate crippling of signals civilians could use until the politicians made it free - one good thing politicians did.
But they don’t make EUV machines today. Think we will still say that 5 years from now? Any gap between their in-country production and what’s blocked by sanctions is going to go away, and probably sooner than later.
Others haven’t been as unimpressed at the pace of Chinese advances as you.
I work in state government. Our computers and servers are usually several years out of date.
Getting money budgeted to replace 20 computers and a file server in a department is difficult. I submit requests that often get rejected.
I suspect it’s similar in military offices. They have to keep compatibility with old legacy mainframe applications. The military’s weapons tech may be state of the art.
I was selected by my federal (civilian) agency to be part of the then Columbia space shuttle recovery. Our base was a collection point for the daily collections and storage search crews found during the day. Every so often artifacts were picked up and transferred to the proper facilities. We saw a lot of artifacts.
Periodically, as new search crews were brought in and trained, they each went through extensive briefings. A point that was hammered into everyone concerned the sensitivity and security of all recovered artifacts.
In every briefing someone asked the question, and the answer was always the same. “Every NASA mission, manned or otherwise, is a military mission.”
What? The first 5nm chip ever produced was in 2017, it entered mass production in 2020. The first 3nm chips were produced in July & December of 2022 by Samsung and TSMC respectively so 5nm has only been “not leading edge” for 18 months at this point and for mass production, only at the launch of the iPhone 15 Pro in September of 2023 so less than 4 months.
There’s some stuff that brags about meeting MIL-STD-810 which is always misleading because 810 isn’t a test, per se. It’s a document that contains hundred of tests that can be tailored to almost any need, and without the specific test, the parameters used for testing and the test results, it’s a useless claim.
When it comes to the names of the chip fab nodes the size in the name has long ceased to match the actual feature size. The tricks used to get the feature size down are insane, and push optics past the bleeding edge.
Fabs using AMSL’s EUV machines are always going to be a generation ahead of DUV. That China’s DUV could be pushed to 5nm equivalent is testament to how smart their engineers are at pushing the tech at their disposal. But it is pretty unlikely they can go further without a wavelength change. And that step to EUV was insane.
But the other question would be whether it really matters. Even when the fabrication tech stops improving, there are lots of things where effort can be applied. We may see aggressive development of novel architectures happening in China. If their government has the will they can fund a lot of research. It isn’t as if they are short of capable scientists. They may hope that they can perform technological leapfrogs in other important areas. Nor is it as if such things haven’t happened before.
Would building up a domestic chip manufacturing capacity in China be a prerequisite for invading Taiwan? If Taiwan’s chip fabs are destroyed, China would still be able to manufacture at least chips for weaponry.
It is really the other way around. China isn’t reliant upon Taiwan’s chip manufacturing - which mostly means TMSC. The west is very reliant. So long as TMSC is the leader of the pack - both in terms of process but also in terms of yield and thus cost, the West remains very nervous about China’s attitude to Taiwan. It is of course more complicated, but TMSC does figure large in the question. Lots of leading edge chips made by TMSC are shipped to China to assemble into products - famously iPhones. But China generally doesn’t have access to arbitrary tech chips from TMSC.
Whether an invasion of Taiwan would result in China acquiring TMSC’s capability is another question. One of the reasons, IMHO, a full scale military invasion is unlikely. China generally plays a long game. Generally.
Current efforts to get leading edge fabs operating outside of Taiwan are part of the west derisking things for themselves. But the labour conditions for fab workers in Taiwan, and the general work ethic and application of workers to the task is not going to be replicated in the west. So the economic advantage that TMSC and Taiwan generally have is going to remain.
So, a generation behind, as I said. And they had access to the technology to build that, as I said, since the restrictions were not yet in place.
There is only one company in the world that can make the most advanced machines, so this is non-trivial. In ten years, sure, but we’ll have moved on by then.
And as the article said, most volume is relatively large and cheaper chips. Bleeding edge stuff is expensive - I worked on it, I know how much wafers cost.
Sometimes political leaders do irrational things. But yes, the main reason for the CHIPs act is to bring production back here. At the moment we’d be screwed whether or not the TSMC fabs got blown up.
I worked with TSMC. They are good. I had code which logged onto their servers about 16 times a day to pull data for our chips. But the people at Intel fabs are also good, and their work ethic is fine. Intel has fallen behind the tech race, but with any luck that gap will be closed. People in the field are very excited about the CHIPs act, since it gives an incentive for fabs to move closer. The real problem is that new process technologies are so expensive that only TSMC and Intel can afford them, and Intel is falling behind because they bet wrong on chips for the mobile world.
I don’t know what labor costs are for a fab, but I can’t imagine they are much as compared to depreciation on the equipment.
