As an ASML US employee, I’ve heard a lot of the internal lore/history about SVG and its predecessor Perkin-Elmer, but it’s interesting to hear more about why Nikon and Canon missed out.
I of course can’t talk about anything confidential, but just to speak in broad strokes, most of the last 20 years of EUV work has been in the reliability space. When I was hired a while back, ASML’s public communications were heavily focused on proving to the customers that we were finally achieving various milestones related to EUV process profitability. Basically, machine uptime was just crossing that threshold barely within the last ten years, for the simple reason that it is insanely difficult.
This is why in my personal opinion I’m skeptical of the scare stories out there warning of Chinese reverse engineering of EUV tech. This is one of the most difficult industrial learning curves that exists. China has a lot of things going for it, but “magic” isn’t one of them.
Not when it pertains to a hostile techno-communist country that openly declares its intention to degrade and destroy the U.S.
All the advanced tech knowledge (including the bioweapon known as Covid) — all because greedy global corporations wanted acces to China's slave-labor manufacturing!
You write that in 1996 "Congress" voted to defund EUV research - and I had to look it up. ASML should thank the Republican House Speaker Newt Gingrich for the assist.
We seem to be forgetting the unsung heroes who helped realize moore’s law: (from the book: “island tinkerers: innovation and transformation in the making of Taiwan’s computing industry” by Honghong Tinn) pg. 189 “I argue that Taiwanese electronics factory women enabled the miniaturization of electronics and mass production of them, in the late 1960s and 1970s. Their assembly of electronics contributed to the realization of moore’s law and made possible offshore production of highly sophisticated electronics for multinational corporations.”
We also seem to be forgetting the boys on the oil rigs, the unsung heroes who helped realize Moore's law. Probably nobody has written a book trying to shoehorn their identity into the story, though.
The history of how Silicon Valley Group could not find committed investors which left ASML the only available rescue is worth looking at. The investment community in the USA simply does not provide money for the long bets. The USA government funded the research side of EUV but did not help with funding production. Intel did fund SVG at key points but the project which stretched SVG too far - the project which became ASML twinscan that won the optical market - was beyond the patience of USA customers or investors.
Intel did not flub the 10 nm node due to betting on optical. In that same time frame TSMC was using optical for the very similar N7 node which was a huge success, no EUV used at all. Intel had other problems in its technology choices and that probably caused them to delay EUV, so you are repeating a story that has the causality backward. Mainland China is pushing optical to roughly N5 equivalent, though one cannot say that it would be commercially successful if it were not necessary due to the embargo on EUV. But it does reinforce that lack of EUV was not the initial cause of Intel's process woes, it was the process problems that delayed them getting EUV.
The gap between developing breakthrough technolgy and successfully comercializing it is such a critical lesson here. Intel bet big on EUV early and funded most of the research, but the US lithography industry had already lost market position to Nikon and Canon. ASML being seen as neutral ground ended up being the deciding factor. Its a reminder that industrial competitiveness isnt just about innovation its about maintaining manufacturing ecosystems.
As an ASML employee I’d like to mention that after the 2012 the viability of EUV for high volume chip production was by no means secure.
ASMLs management basically bet the company on EUVs success despite still having significant, seemingly insurmountable engineering challenges before TSMC even would consider it as an option, such as the light source power and defect-free mask.
ASML was the only player left willing to keep pouring money into EUVs very uncertain future.
ASML is the exemple of integration done right. It developed the hardest parts of the technology itself, build the supply chain (not easy, this is the most complex production machine in the world) and revolutionized some related technologies (ZEISS mirrors are now the most precise surface that exists)
I don't suppose there's a source for Sandia's soft x-ray research (i.e. EUV) coming out of SDIO funding? I see a number of papers jointly published by Sandia and AT&T Bell Labs and a DOE funding number, but not an obvious link to SDIO.
Fascinating history, demonstrating the challenges in technology development from basic science to engineering to manufacturing. Some of it is grueling and seemingly boring. It took a lot of nerve for those companies to invest gargantuan sums into ASML. I wonder if our present system can support such R&D bets (and the losses that come with high-risk, challenging areas).
As an ASML US employee, I’ve heard a lot of the internal lore/history about SVG and its predecessor Perkin-Elmer, but it’s interesting to hear more about why Nikon and Canon missed out.
I of course can’t talk about anything confidential, but just to speak in broad strokes, most of the last 20 years of EUV work has been in the reliability space. When I was hired a while back, ASML’s public communications were heavily focused on proving to the customers that we were finally achieving various milestones related to EUV process profitability. Basically, machine uptime was just crossing that threshold barely within the last ten years, for the simple reason that it is insanely difficult.
This is why in my personal opinion I’m skeptical of the scare stories out there warning of Chinese reverse engineering of EUV tech. This is one of the most difficult industrial learning curves that exists. China has a lot of things going for it, but “magic” isn’t one of them.
I hope you are correct.
Why? More advanced technology in the world is a net positive.
Hoarding it is anti progress
Not when it pertains to a hostile techno-communist country that openly declares its intention to degrade and destroy the U.S.
All the advanced tech knowledge (including the bioweapon known as Covid) — all because greedy global corporations wanted acces to China's slave-labor manufacturing!
A month later and this comment seems like this would describe the US more than China
You write that in 1996 "Congress" voted to defund EUV research - and I had to look it up. ASML should thank the Republican House Speaker Newt Gingrich for the assist.
We seem to be forgetting the unsung heroes who helped realize moore’s law: (from the book: “island tinkerers: innovation and transformation in the making of Taiwan’s computing industry” by Honghong Tinn) pg. 189 “I argue that Taiwanese electronics factory women enabled the miniaturization of electronics and mass production of them, in the late 1960s and 1970s. Their assembly of electronics contributed to the realization of moore’s law and made possible offshore production of highly sophisticated electronics for multinational corporations.”
We also seem to be forgetting the boys on the oil rigs, the unsung heroes who helped realize Moore's law. Probably nobody has written a book trying to shoehorn their identity into the story, though.
Well Ian Kumekawa in his latest book: “empty vessel” mentions the men who worked on oil rigs off the coast of Britain called “thatcher’s children.”
The history of how Silicon Valley Group could not find committed investors which left ASML the only available rescue is worth looking at. The investment community in the USA simply does not provide money for the long bets. The USA government funded the research side of EUV but did not help with funding production. Intel did fund SVG at key points but the project which stretched SVG too far - the project which became ASML twinscan that won the optical market - was beyond the patience of USA customers or investors.
Intel did not flub the 10 nm node due to betting on optical. In that same time frame TSMC was using optical for the very similar N7 node which was a huge success, no EUV used at all. Intel had other problems in its technology choices and that probably caused them to delay EUV, so you are repeating a story that has the causality backward. Mainland China is pushing optical to roughly N5 equivalent, though one cannot say that it would be commercially successful if it were not necessary due to the embargo on EUV. But it does reinforce that lack of EUV was not the initial cause of Intel's process woes, it was the process problems that delayed them getting EUV.
The gap between developing breakthrough technolgy and successfully comercializing it is such a critical lesson here. Intel bet big on EUV early and funded most of the research, but the US lithography industry had already lost market position to Nikon and Canon. ASML being seen as neutral ground ended up being the deciding factor. Its a reminder that industrial competitiveness isnt just about innovation its about maintaining manufacturing ecosystems.
https://krebscycle99.net/2012/02/06/christina-romer-and-manufacturing/
It took me a second to realize this wasn't about Europa Universalis 5.
lol same here
As an ASML employee I’d like to mention that after the 2012 the viability of EUV for high volume chip production was by no means secure.
ASMLs management basically bet the company on EUVs success despite still having significant, seemingly insurmountable engineering challenges before TSMC even would consider it as an option, such as the light source power and defect-free mask.
ASML was the only player left willing to keep pouring money into EUVs very uncertain future.
ASML is the exemple of integration done right. It developed the hardest parts of the technology itself, build the supply chain (not easy, this is the most complex production machine in the world) and revolutionized some related technologies (ZEISS mirrors are now the most precise surface that exists)
I don't suppose there's a source for Sandia's soft x-ray research (i.e. EUV) coming out of SDIO funding? I see a number of papers jointly published by Sandia and AT&T Bell Labs and a DOE funding number, but not an obvious link to SDIO.
what about the role of IMEC ?
Fascinating history, demonstrating the challenges in technology development from basic science to engineering to manufacturing. Some of it is grueling and seemingly boring. It took a lot of nerve for those companies to invest gargantuan sums into ASML. I wonder if our present system can support such R&D bets (and the losses that come with high-risk, challenging areas).
A small fact check - it is Tanya Jewell, not Tania.