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!
The Chinese seem to build their fabs in like a year and a half. By the second year they would be moving tools in. Probably starting trial production. TSMC Japan was built in a similar time scale. Which is quite impressive and shows the high priority the Japanese government gave to the project...
Intel does not have the capital to expend buying expensive EUV tools wantonly. The shells are much cheaper than the equipment of the fab and have huge construction times. Especially in the US.
I think Intel's expectation is that once their first US fabs start mass producing chips they will get...
Catastrophic. And then people think Intel is doing bad.
Samsung clearly did not do enough to attract new customers or fix their yield issues.
I wonder what will happen to IBM's Tellum II in light of this.
Performance might be salvageable with tweaks to the operating system. The OS scheduler seems to be sending workloads to the E-cores more than necessary because of changes to the chip topology. Windows seems to be the worst offender in this regard.
From what I understand Intel does process development in the US. That then that gets replicated (copy exactly) elsewhere. Including Israel.
Intel does conduct R&D in Israel but I think that is for chip design, software, and things like that.
Unfortunately this is a general case in the US economy. Lack of investment is killing not just Intel but also Boeing and other US majors. They are treating the company as a piggy bank they can continuously take money out of. But in the technology business without constant investments you will...
That makes a lot more sense than the Huawei Ascend 910B conversation.
Sophgo was making its own AI chips at TSMC and caught the attention of the US government.
The US have been sanctioning all the Chinese AI companies. Not just Huawei's HiSilicon but also Cambricon and Biren.
If you design HPC...
It could have been he passed information on the status of technological progress happening, not specifics on processes, inside CXMT and that is what got him into trouble.
Right now CXMT is a target for US sanctions on semiconductors, so they definitively don't want the West to know of internal...
Pat was blunt but he said something which needed to be said. And this was not an insult to Taiwan itself. It is just because of who their neighbors are. Nothing to do with Taiwan personally.
Of course then CC Wei is perfectly reasonable in taking that statement as an assault on the viability of...
I think we are going to see a paradigm shift in the DRAM sector with 3D-DRAM replacing the current planar memory.
That will enable huge advances in bit density like we saw with V-NAND.
I think it is still too early to tell. Pat is still fixing lingering issues with underinvestment at Intel. You don't stop building fabs for a decade, with continual tweaks to your now outdated equipment, and then expect turnaround to happen even in four years alone.
The board should give Pat at...
These articles are all rehashes of the same thing. Something is clearly not being properly explained here, even if there is a kernel of truth to the story, the way it is explained makes no sense. So I give it low credibility.
That report is reasonably accurate. But SMIC is building three fabs, not two. At Shanghai, Beijing, and Tianjin.
All the SMIC and Hua Hong fabs are also gigafabs with 100,000 wpm or close to that of capacity.
If the Hua Hong Fab 10 is the one I am thinking about then it still has not started...
That last chart makes it look like SMIC has less capacity than UMC. Totally bogus.
SMIC is #3 in capacity behind TSMC and Samsung. And Hua Hong will surpass UMC and GlobalFoundries to become #4 until the end of the decade.
This is kind of overstated since China already has higher labour prices than Mexico.
But yes 200mm wafer FOUPs can be handled by people much more easily so it decreases the amount of fab automation you need to get a facility up.
Last one they built was a decade ago I think?
It is a bad idea to build a 200mm line as a foundry today. Has been for a long time. Bad economics.
Unless you are talking about niche specialized processes.
The 200mm equipment has likely already been deployed. As much as anyone in the industry would like for there to still be equipment leftover to use somewhere, good luck getting some. But this was not done by any of the major players in China. The 200mm equipment was sent to the small players like...
