Sure looks like Beijing stole blueprints from chip fab world's ASML A former ASML worker accused of stealing trade secrets for advanced chip-making equipment from his employer is now suspected of spying for the Chinese government. Citing sources familiar with an internal ASML probe into the alleged theft, Bloomberg reports the employee, who is apparently based in China, had “potential” ties to …
Warren Buffett Just Dumped TSMC shares.
It was either:
1) Someone knew Taiwan would soon be invaded
2) Someone knew homegrown China EUV tech was soon going to be announced
By this news, I believe it is option 2). Not necessarily due to spying, tough. USA spy China way more.
The New York Times:
C.I.A. Admits to Losing Spies in China
Oct 7, 2022
____
New York Times:
Killing C.I.A. Informants, China Crippled U.S. Spying Operations
May 20, 2018
___
Daily Mail:
CIA admits its spies were executed by Iran and China after communications breech
Oct 5, 2022
____
The Hill:
CIA admits to losing dozens of spies around the world
Oct 5, 2021
____
Business Insider:
How China Found CIA Spies and Executed Them
Aug 16, 2018
____
Foreign Policy:
China's Secret War for U.S. Data Blew American Spies' Cover
Dec 21, 2021
____
Foreign Policy:
Botched CIA Communications System Helped Blow Cover of Chinese Spies working for CIA
Aug 15, 2019
____
The Guardian:
China 'dismantled' CIA spying operations and killed Spies
May 20, 2018
Yeah. Obviously it would be an error to assume the number of productive researchers in a nation is exactly proportional to its population, given the large disparities around the world in education, opportunity, intellectual freedom, and so forth. But China has a ton of people, and a ton of universities and corporate R&D departments. It's absurd to believe there isn't a great deal of primary research happening there, and of course many Chinese researchers publish research, so there's plenty of empirical support too.
This was guaranteed to happen. The Dutch intelligence services are pretty competent (remember when they had hacked into the security webcams of the Russian troll farm that was trying to influence US elections?) and I’m sure they have intercepted far more attempts that we don’t know about.
One of the Bloomberg reporters is Jordan Robertson. He published an obviously fake news story about a china manipulated Supermicro server four years ago. The story is still up om Bloomberg without a correction.
Robertson has published 36 stories about cybersec since the fake China story.
He even has it nailed on top of his Twitter feed with no comment.
https://twitter.com/jordanr1000
Noting particularly sus about the ASML spy story per se.
But a strange choice by Bloomberg to keep using Jordan Robertson and not in some form retracts the old story!
This would be like if I stole blueprints for a nuclear reactor. Thanks to the blueprints I might know how to put all the parts together to build one, but even if I had an unlimited budget and ignoring stuff like getting fuel rods having all the required parts made out of the exact right alloys and to the necessary tolerances would take years before I had something that worked half as well as a proper reactor.
Now multiply that difficulty by 100, because the tolerances of of a modern EUV scanner are in the microns and nanometers, the materials are far more exotic and in many many cases have only ONE source in the entire world, and when that one source is based in the US, EU or Japan as most are they will not sell to China.
It's probably not an ACTUAL blueprint. So the data will also specify materials, parts, etc.
And China has the budget to do almost anything. Can you back up your statements about the rarity and non-substitutability of the parts used?
There are supporting technologies and machines that are needed, not just the imaging machine. Making the photomasks has been a major issue to overcome. A number of the supporting tech was developed in Japan as part of their EUV attempts.
It won't specify materials and parts because ASML does not make those, they get them from other suppliers. You'd need to steal information from hundreds of companies to be able to build an EUV scanner and then figure out how to duplicate manufacturing of something that often required them years to perfect.
Indeed. The key invention that enabled them to do what they do is the light source. The rest will, of course, also need some supply chain creativity/spying, but the core is the way they generate the light so I assume that's where they will focus their efforts.
Not just being able to generate it. Inside the device there are at least a dozen mirrors that reflect the EUV light to and fro to focus it at the reticle mask. Figuring out how to make mirrors that reflect EUV was not easy I imagine, as they contain over 100 layers of various materials have to be flat to within a single atom. The example ASML gave was that if the mirror was the size of Germany the tallest mountain would be 1mm in height.
There is a LOT of technology that has been worked on for at least two decades that finally came together to make production volumes of EUV chips economic to produce. China won't get there quickly, though they can probably obtain some limited capability for producing chips matching TSMC's leading edge before the end of the decade if they are willing to make sacrifices in terms of how economically they can be produced and in what numbers. For military applications being limited to only producing a few wafers per day and chips costing $50K each might be fine, but obviously that won't help companies like Huawei compete in consumer or corporate markets.
Its at least possible they could make a breakthrough in much higher NA DUV lenses or somehow make ebeam economic at mass production volumes, so I doubt they will put on all their eggs into trying to copy ASML's EUV technology.
China has long had a reasonably capable optical industry producing, for example, lasers and high efficiency dielectric mirrors, both of which you *really* need for EUV.
While pretty much every part of an EUV-based system is eye-wateringly hard to design, build and run , I can't really see why Chinese engineers wouldn't eventually get there.
That being the case, you'd imagine that any little 'how to' tidbits half-inched from ASML might be useful.
When Nokia was king cheap knock-offs from Shenzhen with all the latest features duplicated would appear within three weeks at Sing Tat Plaza in Mongkok, Hong Kong.
Nowhere near cutting edge semiconductor technology but it was a start and that was 20 years ago.
Today Shenzhen has Huawei.
The crafty element is how ASML produces the light for exposure, that took decades to develop to a level that it could be used in production. I don't think the rest would pose many challenges once they had the details for that, so I suspect that's the exact thing that ASML will protect as their crown jewels. Because it's one specific component it's relatively easy to containerise that data and protect it, but the usual security imbalance exists which is never in favour of the defenders: they only have to fail once :(.
That said, I would lace the place with fakes.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
