Chinese semiconductor industry

Status
Not open for further replies.

tokenanalyst

Brigadier
Registered Member
looks like a new hhgrace bid won by SMEE.

Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. Announcement on the Winning Bid Result of the 12-inch Integrated Circuit Manufacturing Project (1)​

  • Project number: 0705-234023603809
  • Announcement Type: Result Announcement

Project name: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. 12-inch integrated circuit manufacturing project
Project number: 0705-234023603809
Bidding scope: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. 12-inch integrated circuit manufacturing project
Tendering agency: Shanghai International Tendering Co., Ltd.
Tenderer: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd.
Bid opening time: 2023-09-25 09:30
Announcement time: 2023-10-07 13:02 - 2023-10-10 23:59
Bid-winning result announcement time: 2023- 10-11 19:10
Winning bidder: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer country or region: China
 
Last edited:

tokenanalyst

Brigadier
Registered Member

High-precision lithography machine bidding result announcement (1)​



Project name: High-precision lithography machine
Project number: 0664-2340SUMECF85/06
Bidding scope: High-precision lithography machine
Bidding agency: SUMEC International Technology Trading Co., Ltd.
Tenderer: Zhuhai Tiancheng Advanced Semiconductor Technology Co., Ltd.
Bid opening time: 2023-09- 07 10:00
Announcement time: 2023-09-12 17:19 - 2023-09-15 23:59
Bid result announcement time: 2023-09-18 15:50
Winning bidder: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer country or region: China
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
looks like a new hhgrace bid won by SMEE.

Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. Announcement on the Winning Bid Result of the 12-inch Integrated Circuit Manufacturing Project (1)​

  • Project number: 0705-234023603809
  • Announcement Type: Result Announcement

Project name: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. 12-inch integrated circuit manufacturing project
Project number: 0705-234023603809
Bidding scope: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd. 12-inch integrated circuit manufacturing project
Tendering agency: Shanghai International Tendering Co., Ltd.
Tenderer: Huahong Semiconductor Manufacturing (Wuxi) Co., Ltd.
Bid opening time: 2023-09-25 09:30
Announcement time: 2023-10-07 13:02 - 2023-10-10 23:59
Bid-winning result announcement time: 2023- 10-11 19:10
Winning bidder: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer: Shanghai Microelectronics Equipment (Group) Co., Ltd.
Manufacturer country or region: China
yep, I didn't post the picture that a friend sent me, but looks like it happened. Although, I would say this is most likely Arf dry, since it is at the Wuxi fab
 

LanceD23

New Member
Registered Member
Looking at the numbers, in particular on the difference between Q2 and Q3 on China market, SMIC moved from 754K wpm capacity at 78,3% utilization with 79,6% of which sold to China, to 796K capacity with 77,1% utilization and 84% sold to China.

This means that volume sold to china increased from 470K wpm in Q2 to 515K wpm in Q3, this is a +10% QoQ on domestic market. This is not a small increment.

Assuming that in case of further demand and further decopuling from US customers, SMIC can realistically grow to 90% utilization with 90% on domestic market, this at current capacity equals to 644K wpm, i.e. about +130K wpm of new available volume for China market, even without capacity expansion.
question from Newbie.
Huwei's projected volume , maybe 100M+ chips next year? (cellphone, tablet, AI, EV)
SMIC can handle it?
 

tonyget

Senior Member
Registered Member
I'm not familiar with HBM. But how similar or different,between the hybrid bonding used on 3D NAND and HBM?

Please, Log in or Register to view URLs content!

If China cannot import HBM, it would have to develop its own. Despite media reports last summer that China’s DRAM vendor ChangXin Memory Technologies (CXMT) is developing an “HBM-like” DRAM, Lorenz believes it’s still a long way off.

He added that some of the underlying technologies include hybrid bonding, under development at Yangtze Memory Technologies Corp (YMTC). That’s not the same as HBM, but China is “at least building the knowledge foundation behind these packaging technologies” which could help in the long run, he added.

Asked about future Chinese chiplet designs lacking HBM access, Ron Wilson, technology editor for the Ojo-Yoshida Report, noted that developing HBM-like DRAM is challenging, adding, “This should not require any technology that would violate U.S. export controls.”

Wilson continued, “Maybe the result wouldn’t be quite as dense or as fast as what Micron or Samsung could do, but it should be sufficient to keep China in the AI game. The same is true for the stacking and interposer [or organic substrate] technology necessary to combine chiplets with HBM stacks.”
 

hvpc

Junior Member
Registered Member
I'm not familiar with HBM. But how similar or different,between the hybrid bonding used on 3D NAND and HBM?

Please, Log in or Register to view URLs content!
Hybrid bonding connects chip-to-wafer or wafer-to-wafer via Copper Pads on the backside of each chips.

HBM uses through-silicon-vias (TSV) & micro-bumps to connect DRAM chips.

Stacking or bonding technique used in HBM today is not the same as hybrid bonding. No similarities come to mind. But, as the HBM tiers grow to 12-tiers next, we may migrate to using hybrid bonding instead of TSV/uBump.

The technology behind HBM is not difficult. But with the current DRAM capability of CXMT, stacking those chips may give you HBM2 level capability. But I don't see any point in using CPU/GPU that's a few generations behind (~1/3 the transistor density) along with HBM that's also a few generations behind (~1/4 the bandwidth & I/O speed) to come up with domestic GPU alternatives to compete against Nvidia. The gap will only continue to grow if we are stuck at N+2 & 18nm DRAM.
 
Last edited:

tonyget

Senior Member
Registered Member
The technology behind HBM is not difficult. But with the current DRAM capability of CXMT, stacking those chips may give you HBM2 level capability. But I don't see any point in using CPU/GPU that's a few generations behind (~1/3 the transistor density) along with HBM that's also a few generations behind (~1/4 the bandwidth & I/O speed) to come up with domestic GPU alternatives to compete against Nvidia. The gap will only continue to grow if we are stuck at N+2 & 18nm DRAM.

What about third party companies buy DRAM from CXMT,then stack up the chip themself?I have seen quite a few Chinese companies entering this space

I agree that advanced node process,is something China can not bypass. For logic chip and DRAM,advanced node will always matter.
 
Last edited:

hvpc

Junior Member
Registered Member
What about third party companies buy DRAM from CXMT,then stack up the chip themself?I have seen quite a few Chinese companies entering this space
Like I said earlier,, TSV & uBump are not difficult and many BE fabs have such technologies. But I don't think they can't just buy off-the-shelf DRAM chip. Don't quote me on this, but I think the die layout of DRAM chips used for HBM are different from the "regular" ones. Either that or the backend metal routings are different. In any case, CXMT and the BE fab would need to work together.
 
Status
Not open for further replies.
Top