Comment

Comment on Taiwan refuses to move half of U.S.-bound chip production to American shores — trade discussion to be focused on Section 232 investigation for preferential deal on semiconductors

<- View Parent

UnderpantsWeevil@lemmy.world ⁨3⁩ ⁨weeks⁩ ago

Good. It was a naked attempt at a shakedown, and also 100% a bluff.

It takes more than a few years to spin up a chip fab, with an outlay on the order of hundreds of billions to trillions of dollars. Even if they’d been eager to take the US up on the deal (and why would they want to relinquish a functional monopoly on cutting end processors?), there’s no way they’d be dealing with the same administration by the time it was completed. Even if Trump was still in office, the fucker changes his mind every five minutes. Not conducive to long-term economic projects like this.

Now is emphatically not the time to undermine their strategic defense policy, which largely revolves around “if the CCP invades, we will melt our chip fabs to slag”.

TSMC won’t have their edge forever. China’s fabs are catching up quickly, with 5nm chips in production and 3nm chips possible in a few more years. This was a good strategy when China needed to import these chips and Taiwan had the market cornered. But if TSMC’s rigged-to-explode labs go up in smoke after China’s a major player in the market, that actually benefits Beijing.

Strapping yourself with Semtex might be a savvy play in a single moment, but it’s not going to work long term.

That’s before you consider the real threat Taiwan poses to China is as a launchpad for US strikes into the interior.

source

Sort:hotnew top

gravitas_deficiency@sh.itjust.works ⁨3⁩ ⁨weeks⁩ ago
Oh I know - but I’m saying they should halt efforts now, because they’ve been going on for several years (I think close to 4-5?) at this point.

I was also under the impression that Mainland was still meaningfully behind the cutting edge, that TSMC was absolutely not resting on their laurels, and that the prospect of the CCP fabs fully catching up isn’t super likely. Out of curiosity, do you have any references/articles about recent CSMC/etc lithography advancements?

source
- UnderpantsWeevil@lemmy.world ⁨3⁩ ⁨weeks⁩ ago
  The Chinese state industries have been happy enough to throw big chunks of their GDP at the problem of high end chip fab, and it’s paying dividends.
  
  That’s not to say TSMC is idle, but the whole problem of living on the bleeding edge is that you’ve got nobody to crib from. All your next-gen advances have to be earned through high end R&D and brute force engineering and lots of money and time. Their rivals can reverse engineer their technology, learn from TSMC’s mistakes, and generally coast in their wake.
  
  What’s sort of incredibly about America’s Intel is that they haven’t done any of this shit, clinging to their dead-end chip design long after its expiration date and missing the boom in demand for high end chips entirely.
  
  Out of curiosity, do you have any references/articles about recent CSMC/etc lithography advancements?
  
  Nothing you couldn’t just Google up yourself, I’m sure. I picked up SMIC based on their advances in DUV lithography employed by ASML and it paid out big. The high margins on sale are justifying comparatively lower success rates of manufacturing.
  
  source
  - dogslayeggs@lemmy.world ⁨3⁩ ⁨weeks⁩ ago
    
    What’s sort of incredibly about America’s Intel is that they haven’t done any of this shit, clinging to their dead-end chip design long after its expiration date and missing the boom in demand for high end chips entirely.
    
    This is the most baffling thing to me. How could Intel leadership be so incompetent? They had the inside track to hundreds of billions in revenue and just decided to coast.
    
    source
    BakerBagel@midwest.social ⁨3⁩ ⁨weeks⁩ ago
    Because modern western business is about cashing out the business to squeeze out a couple extra bucks each quarter. They aren’t interested in making a product, just cutting costs, raising prices, and issuing stock buybacks
    
    source
    UnderpantsWeevil@lemmy.world ⁨3⁩ ⁨weeks⁩ ago
    
    How could Intel leadership be so incompetent?
    
    For the leadership, it was a cash cow. They got a fat dividend doing very little, even as upstarts blasted past them.
    
    They had the inside track to hundreds of billions in revenue and just decided to coast.
    
    At some point, the effort to get from $10B to $100B isn’t worth the pressure. How many extra yachts do I actually need?
    
    source
    -> View More Comments
    LemmyThinkAboutIt@lemmy.zip ⁨2⁩ ⁨weeks⁩ ago
    If I remember correctly, but it’s been awhile since I read the discussion about Intel so I could be misremembering, the problem with Intel started when the C-Suite stopped being engineers that moved up in the company.
    
    source
  - gravitas_deficiency@sh.itjust.works ⁨3⁩ ⁨weeks⁩ ago
    Interesting, thanks. And yeah, I too find it utterly baffling at how Intel is turning into a has-been before our eyes. They were The Chip Guys for ages, and then the fucking quants got put in charge and carved away so much of the engineering leadership and underpinnings that it’s a husk of what it was in the 80s and 90s.
    
    I would, however, point out that TSMC’s whole deal is “define, and produce at scale, the bleeding edge of integrated circuit designs”, so the bit about them cribbing off of people hasn’t really been a variable in their equations for at least a couple decades. They have been major (arguably, the predominant) pioneers in chip lithography for a while now.
    
    source
    UnderpantsWeevil@lemmy.world ⁨3⁩ ⁨weeks⁩ ago
    
    so the bit about them cribbing off of people hasn’t really been a variable in their equations for at least a couple decades
    
    That’s meant to say they can’t do that from the front of the pack. There’s not much to crib from that they don’t invent.
    
    That’s a burden on TMSC that guys like Samsung and SMIC don’t suffer from, at least until they can match pace. For the time being, other firms can close in on TMSC by following in their footsteps (or, at least, avoiding their errors).
    
    source
    -> View More Comments
  - Rentlar@lemmy.ca ⁨3⁩ ⁨weeks⁩ ago
    Intel: did I hear somebody say we should announce our newest 14nm+++++++++ platform?
    
    source
    humanspiral@lemmy.ca ⁨2⁩ ⁨weeks⁩ ago
    14a =1.4nm. Given that they cancelled 18a to leap frog to 14a is not a confidence builder.
    
    source
humanspiral@lemmy.ca ⁨2⁩ ⁨weeks⁩ ago
China has the engineering talent and numbers to catch up for certain, but SMIC is not producing advanced chips in volumes, even as designs for next gens of chips and SMIC technology come out fast.

The big factor in all of this is that the market for chips in China is 5x+ that of US, and the business interests of anyone in the sector outside of China would be to choose China over US if they only had to pick one. US IP is going to expire soon enough, but is already abused for colonial power over global chip sector.

China is definitely at a mature point where home grown chips can already compete in phones/laptops (Huawei) and AI due to their energy infrastructure. They don’t need to invade Taiwan to have useful electronics, and the home grown industry will accelerate faster than West’s. While TSMC’s margin of leadership will narrow, they will still be ahead for more than 5 years.

source