You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
Told ya. More customers for SMIC. Thanks Uncle Sam.
The market is already factoring in the future revenue growth for the Chinese semi manufacturers.
In a year or two those chips will be redesigned from a TSMC to a SMIC process. Just like what happened to Huawei's HiSilicon or Phytium. And due...
The RISC-V Consortium has been putting out standards on IOMMU, ACPI, and other system components. It is way more standardized than ARM was at this point in development. The reason Linus disliked the ARM environment was because each implementation was pretty much its own custom thing. The ARM ISA...
Isn't that the Japanese guy who bought WeWork and is known for making ruinous malinvestments?
No wonder he failed to buy NVIDIA before the stock skyrocketed.
For what it's worth AMD tried to push the ATI engineers post acquisition to use GlobalFoundries instead of TSMC. Initially those engineers did not want to do it because the EDA environment for GlobalFoundries was way less developed. Then GlobalFoundries started fumbling continuously delaying the...
Outsource more what? Arrow Lake is made at TSMC and that is that.
I find it surprising Intel would want to increase orders of it considering the launch flopped.
There were claims 20A would be used for Arrow Lake at one time but with the cancellation of mass production in that node TSMC is the...
Hwatsing, NAURA, ACM Research, in that chart are Chinese companies. The Chinese fab equipment supplier market is way more fragmented in terms of suppliers so I bet a lot of them are in the gray area since the chart only shows the top four in each segment. For example AMEC is a fairly large...
AMD also had the Spansion NOR Flash partnership with Fujitsu. Which came with its own fabrication facilities.
You could say AMD had half the logic fab capacity they needed at one point. Because of financial issues they never had more than two logic fabs (Austin, Texas and Dresden, Germany)...
AMD was the only competitor to Intel which still had its own fabs. When AMD sold their fabs that should have been a wake up call for anyone in the industry to figure out the future was the fabless model. While I am of the opinion that Intel should presently continue its IDM 2.0 strategy, which...
I agree that his point of view is ridiculous. As a result of Rock's Law the cost to build a fab will continue to increase which will make the IDM model increasingly untenable vs the Foundry model. In the 1980s a lot of computer manufacturers had their own manufacturing facilities operating as...
No. What he means is that the US will get the 2nm technology after Taiwan has the node after that. Foreign fabs will always be behind.
I already said as much here. TSMC is to retain its most advanced node in Taiwan. And this is expected to be in perpetuity.
If it scales that bad and interconnect is such a bottleneck why are the AI hyperscalers using clusters and connecting machines together? It is just matrix multiplication with short number formats.
Spending $100 billion USD building an AI server farm that will be obsolete in two process generations as a national development project is retarded. And you do not need that kind of money to develop AI models tuned for Arabic.
Back then Intel were running high on sales of PCs during the pandemic so they had the money. And it is not like those 5 processes in 4 years develop themselves.
Chip development is happening much faster than previously. For example you had Intel release Meteor Lake in Q4 2023 and Arrow Lake in...
