[HN Gopher] TSMC told US of chip in Huawei product after TechIns...
___________________________________________________________________
TSMC told US of chip in Huawei product after TechInsights finding,
source says
Author : mikhael
Score : 40 points
Date : 2024-10-22 19:48 UTC (3 hours ago)
(HTM) web link (www.reuters.com)
(TXT) w3m dump (www.reuters.com)
| m3kw9 wrote:
| I'm sure there is some unscrubbable serial number to at least
| trace the first, second merchant
| j_walter wrote:
| There is no serial number on the individual die, but TSMC has
| very unique test lines on the chip that would easily identify
| if it was in fact made at TSMC.
| 0xTJ wrote:
| Is it indicated somewhere in this article, or somewhere else,
| that the dies are not programmed with a unique serial while
| still as part of a whole wafer?
| j_walter wrote:
| No, but any programming that would occur with identifiable
| information would occur after the wafer is cut and
| packaged. Generally this would be after delivery to TSMC's
| customer.
| wiml wrote:
| By default I'd assume most chips don't get serialized
| (their package may have a date/lot code). The article
| doesn't say what the chip is, so I don't think there's a
| reason to think that it's one of the types of chip that
| does usually get a unique ID.
| londons_explore wrote:
| I don't have insider knowledge... But I could believe that
| there is no per-die serial number unless the customer wants
| one. It would be extra manufacturing steps, and require
| integration with the customers design.
|
| Due to the way lithography works, it isn't easy to make
| each die different. The usual way to put serial numbers
| into chips is with efuses, but not all chips have any
| efuses at all, and it would require collaboration with the
| customer to design a way that they be programmed and read
| (probably on a JTAG chain).
| DiogenesKynikos wrote:
| These chips could have been manufactured by TSMC before the US
| issued an export ban:
|
| > It is unclear how the chip made its way to Huawei. In 2019, the
| company released its Ascend 910 chip series. At the time, prior
| to export controls, the chips were produced by TSMC, two sources
| told Reuters earlier this year.
|
| The question of why the US has the right or power to tell TSMC, a
| Taiwanese company, who it is allowed to do business often comes
| up in these discussions. I've often seen the response that this
| is US technology, and that any country would apply similar
| controls to its own technologies. What I don't think people
| realize is that these sorts of "secondary" controls are very
| unusual, internationally.
|
| The US imposes controls on goods manufactured abroad using US-
| made tools or intellectual property. This is a bit like the way
| that the GPL "infects" other projects, and forces them to abide
| by its terms, and to my knowledge, the US is the only country
| that does this (in any case, it's the only country doing this on
| such a large scale). If you think of how integrated the world
| economy is, these sorts of "infectious" controls are extremely
| disruptive.
| burnte wrote:
| They say "if you want to do business in the US, with US
| companies, don't deal with these people/groups/countries."
| j_walter wrote:
| Exactly this...and since the US is by far the biggest revenue
| source for semiconductors it would be stupid to ignore this
| rule. Taiwan in general wants to maintain favor with the US
| because of the invasion threat of China as well (although I'm
| not 100% confident the US would get involved in that fight).
| pessimizer wrote:
| > The question of why the US has the right or power to tell
| TSMC, a Taiwanese company, who it is allowed to do business
| often comes up in these discussions.
|
| The US doesn't have to justify what it asks for, it does that
| as a convenience for Taiwanese politicians. _Why_ it has the
| power is a historical question irrelevant to _whether_ it has
| the power. _Why_ it has the right? Rights are implemented with
| power.
|
| Proving that the US is wrong according to some moral or
| rational standard is just an intellectual exercise.
| nomel wrote:
| > Why it has the power is a historical question
|
| I don't think history is necessary. With Apple alone being
| around 25% of their income [1], them losing a fraction of the
| money from the US would almost certainly be worse than losing
| whomever the US doesn't like at the moment.
|
| [1] https://focustaiwan.tw/business/202303030016
| throw310822 wrote:
| Hmm, what if TSMC just told the US, "ok, go screw yourself,
| I'll sell my services _to anyone but you and your
| companies_? Yes, TSMC would lose a lot. But arguably the
| damage to the US would be a lot more.
|
| But of course ultimately the US has guns and TSMC doesn't,
| and the US wouldn't accept this type of outcome.
| Aloha wrote:
| export compliance is a total absolute tire fire, its full of non-
| sensical policy, and contrary opinions, and the more countries
| your product goes thru, the worse it gets.
|
| Have a part made in Japan, integrated into a product in the
| states but sold by someone in the UK to in France? you'll have to
| comply with Japan, US and UK laws.
|
| Neat fact, the UK considers the Cisco C9200 switch to be a
| munition, because it has ipsec.
| ArchOversight wrote:
| It's not only the UK that considers cryptography to be a
| munition. It is also classified as munitions in the US:
|
| > Encryption items specifically designed, developed,
| configured, adapted or modified for military applications
| (including command, control and intelligence applications) are
| controlled by the Department of State on the United States
| Munitions List.
|
| It was part of the whole crypto wars, and the lawsuit brought
| by Bernstein vs the United States.
|
| See more:
|
| - https://en.wikipedia.org/wiki/Bernstein_v._United_States -
| https://en.wikipedia.org/wiki/Crypto_Wars -
| https://en.wikipedia.org/wiki/Export_of_cryptography_from_th...
___________________________________________________________________
(page generated 2024-10-22 23:00 UTC)