the first High-NA machine in the world (~$300m+) has been delivered to intel by asml
>10 years in the making
>probably 25-30 years ahead of any potential competitors (there aren't any)
>moore's law confirmed not dead
0.55 numerical aperture where old EUV (0.33 na) could print ~13nm features, that is now down to 8nm. that translates to 2-3 times more transistors per mm
>takes 3 boeings to ship it
>yeets silicon at 32g into a mirror wall to print next-gen cpu's
>185 wafers an hour
>depth of focus is like 10nm lmao
>industry scrambling to invent new materials/masks/inspection tools
tldr; the worlds most advanced tech has been shipped and is already in use by intel

>>105981979 (OP)
>Intel
What a waste of human achievement, wish AMD hadn't killed off GlobalFoundries completely.

>>105981979 (OP)
Great, Now all they have to do is design a processor worth making on that process.

>>105981999
global foundries was never going to be in this game, they exited bleeding-edge nodes years ago
intel is the only IDM with enough scale and funds to pioneer high-na
amd still wins from this via tsmc
this isn't about intel really, it's about who owns the future of semiconductors and rn that's asml and whoever can afford asml machines

>>105982054
>global foundries was never going to be in this game, they exited bleeding-edge nodes years ago
They had the technical knowledge but finances were fucked up and AMD jumped ship to TSMC.

https://www.anandtech.com/show/13277/globalfoundries-stops-all-7nm-development

>>105982103
yeah, GF had the know-how in powerpoint form, they even started 7nm dev but what they didn't have was the capital to burn on EUV toolchains, masks, and ecosystem support. they noped out
amd made the right call jumping to tsmc otherwise they'd still be stuck at 14nm

>>105981979 (OP)
>obligatory
Usecase?

Being a brown aside, 3x transistor density does sound promising. Does TSMC have any of these machines yet, or are they still using these NEXTSCAN NXEs?

>>105982167
In hindsight both made the right call but somehow the litography romantic in me wonder what GF could have pulled off if they really tried.

>>105982168
Tsmc is skipping this gen... Its going to be ~2029ish sadly

Even with unlimited resources you'd need 15 years or more to catch up to ASML. If I was the USA or China I'd probably start doing so.

Wow it's worth like one b-tier AI researcher

>>105981979 (OP)
>that translates to 2-3 times more transistors per mm
=2.64x
Or theoretical cost reduction per transistor of 62% (as if)

>>105982332
It's just not ASML, it's also Zeiss and all the other suppliers. They have exclusivity deals with ASML or are owned by ASML, like Cymer.

>>105982209
That seems like a poor decision for them, they can’t do a capital raise with their $1T mcap for some machines to ensure their continuing lead? Definitely too investoor-brained

>>105982041
Their CPUs aren't that bad. They can compete with AMD.

>>105982465
lol, lmao even

>>105982445
>That seems like a poor decision for them
Not if there is no serious competition, and they clearly believe that to be the case.

I don't understand why the chinks can't simply copy them. Surely they have enough spies at ASML, and it's not like they care about patents.

>>105981979 (OP)
>to intel
to the graveyard.
RIP machine.

>>105984251
ASML is mostly an assembler and small clique of designers. You can't simply spy how a society makes the people that constitutes that kind of companies and associates (Zeiss), not even the US could have all the pieces for their ASML, they tried it, even with direct financing by the government but it couldn't keep with the pace of ASML or Nikon. The US in part controls ASML because they have extra leverage after allowing the acquisition of SVG and Cymer (EUV light sources) by ASML. With Cymer and its patents the US could exclude Nikon from EUV, an objective since the late 1980s.

>Surely they have enough spies at ASML, and it's not like they care about patents.
20 years ago they laughed at Changs becaue they said "even if we give your the schematics and instruction you wouldn't be able to make anything functional"
Now...

>>105984251
they're doing their best but speedrunning the whole photolithography supply chain and RnD by yourself alone takes time and money. either way i think the world would benefit from competition.