TORONTO — There’s been lots of chatter about China’s ambitions to become more semiconductor self-sufficient in recent years. The country’s “Made in China 2025” initiative is bearing fruit, with a homegrown company demonstrating further success with its proprietary architecture for 3D NAND flash memory chips. Yangtze Memory Technologies Co.’s (YMTC) 3D 64L Xtacking TLC NAND device is already drawing considerable analysis by Ottawa, Canada-based TechInsights, which specializes in the reverse-engineering of semiconductors in advanced technology products. In a telephone interview with EE Times, senior technical fellow Jeongdong Choe said YMTC’s device will disrupt the $52 billion NAND memory market and its big players such as Micron, Samsung, SK Hynix, Kioxia, Western Digital, and Intel — not only by being a new entrant but because of its Xtacking architecture. “It’s very unique and it’s different from the others. ” Among TechInsight’s early observations are that YMTC’s 3D 64L Xtacking TLC NAND devices comprise 73 gates in total for a vertical NAND string with nine vertical channel (VC) holes, including one dummy hole between common source line contacts. Choe said it’s likely that five select gates (one SSG and four DSGs) and four dummy gates are also used. In YMTC’s proprietary Xtacking architecture, he said, periphery circuits and memory cell operations are processed on a separate wafer, that allows for higher array efficiency and memory bit density than conventional 3D NAND solutions, such as the 64-layer BiCS NAND from Western Digital and Kioxia and Samsung’s 64-layer V-NAND. For example, said Choe, the YMTC 256 Gb die bit density is 4.41 Gb/mm, which is higher than the Samsung equivalent 256 Gb die at 3.42 Gb/mm. YMTC’s Xtacking is quite similar to Micron’s, but distinguishes itself overall by using wafer-to-wafer hybrid bonding, which means the NAND array is upside-down on periphery circuits. As different as Xtacking architecture is from the incumbent NAND players, until quite recently YMTC was still considered to be lagging behind other companies which had already unveiled 96-layer NAND, even 128-layer. However, YMTC recently announced it has developed the X2-6070, a 128-layer 1.33 Tb QLC 3D NAND memory chip, based on its Xtacking architecture. The company also has plans for a 128-layer 512 Gb TLC chip. There’s been a lot of speculation about when China would be able to make an impact in the semiconductor market, as well as the quality of the technology itself. As Jim Handy, principal analyst at Objective Analysis told EE Times last fall, it won’t be the country’s lack of funds that get in the way so much as its lack of experience and sophistication in semiconductor manufacturing. The research released its annual, deep dive report on the factors affecting China’s memory ambitions, including comparisons to investments in other industries in the country and how other countries have kickstarted market entry, such as Korea and Japan. As noted by the Objective Analysis report, Asia Pacific’s sales have clearly grown to account for more than half of world semiconductor sales as all other regions have decreased over the past four decades; China accounts for a growing portion of the region. Less than 20% of the chips consumed in China in 2015 were domestically produced, and the “Made in China 2025” plan is to reach to reach 40% by 2020 and 70% by 2025. YMTC was founded in 2015 and is part of the Tsinghua Unigroup. It designed and manufactured the first 3D NAND flash chip in China in 2017.
这个和DRAM一起可是韩国三星的利润根源所在, 不夸张的说这是韩国的国本. 每年一两百亿美元的纯利润.
https://www.eetimes.com/chinas-memory-ambitions-yield-competitive-nand/#
China’s Memory Ambitions Yield Competitive NAND
TORONTO — There’s been lots of chatter about China’s ambitions to become more semiconductor self-sufficient in recent years. The country’s “Made in China 2025” initiative is bearing fruit, with a homegrown company demonstrating further success with its proprietary architecture for 3D NAND flash memory chips.
Yangtze Memory Technologies Co.’s (YMTC) 3D 64L Xtacking TLC NAND device is already drawing considerable analysis by Ottawa, Canada-based TechInsights, which specializes in the reverse-engineering of semiconductors in advanced technology products. In a telephone interview with EE Times, senior technical fellow Jeongdong Choe said YMTC’s device will disrupt the $52 billion NAND memory market and its big players such as Micron, Samsung, SK Hynix, Kioxia, Western Digital, and Intel — not only by being a new entrant but because of its Xtacking architecture. “It’s very unique and it’s different from the others.
”
Among TechInsight’s early observations are that YMTC’s 3D 64L Xtacking TLC NAND devices comprise 73 gates in total for a vertical NAND string with nine vertical channel (VC) holes, including one dummy hole between common source line contacts. Choe said it’s likely that five select gates (one SSG and four DSGs) and four dummy gates are also used. In YMTC’s proprietary Xtacking architecture, he said, periphery circuits and memory cell operations are processed on a separate wafer, that allows for higher array efficiency and memory bit density than conventional 3D NAND solutions, such as the 64-layer BiCS NAND from Western Digital and Kioxia and Samsung’s 64-layer V-NAND. For example, said Choe, the YMTC 256 Gb die bit density is 4.41 Gb/mm, which is higher than the Samsung equivalent 256 Gb die at 3.42 Gb/mm. YMTC’s Xtacking is quite similar to Micron’s, but distinguishes itself overall by using wafer-to-wafer hybrid bonding, which means the NAND array is upside-down on periphery circuits. As different as Xtacking architecture is from the incumbent NAND players, until quite recently YMTC was still considered to be lagging behind other companies which had already unveiled 96-layer NAND, even 128-layer. However, YMTC recently announced it has developed the X2-6070, a 128-layer 1.33 Tb QLC 3D NAND memory chip, based on its Xtacking architecture. The company also has plans for a 128-layer 512 Gb TLC chip.
There’s been a lot of speculation about when China would be able to make an impact in the semiconductor market, as well as the quality of the technology itself. As Jim Handy, principal analyst at Objective Analysis told EE Times last fall, it won’t be the country’s lack of funds that get in the way so much as its lack of experience and sophistication in semiconductor manufacturing. The research released its annual, deep dive report on the factors affecting China’s memory ambitions, including comparisons to investments in other industries in the country and how other countries have kickstarted market entry, such as Korea and Japan. As noted by the Objective Analysis report, Asia Pacific’s sales have clearly grown to account for more than half of world semiconductor sales as all other regions have decreased over the past four decades; China accounts for a growing portion of the region. Less than 20% of the chips consumed in China in 2015 were domestically produced, and the “Made in China 2025” plan is to reach to reach 40% by 2020 and 70% by 2025. YMTC was founded in 2015 and is part of the Tsinghua Unigroup. It designed and manufactured the first 3D NAND flash chip in China in 2017.
🔥 最新回帖
因为半导体行业有很大周期性，他希望通过兼并那些与本行业无关/ correlation小的企业，来拉平这种周期性。。让自身的财务报表更稳定。。
财务报表稳定 / 现金流没有很大波动的企业，股市估值会更高，也更容易给Hock Tan这种并购狂人来玩金融杠杆。。
个人不是很看好这种方式。。
真要diversify，我作为投资者，自己去买两家不同企业的股票就好了，要他瞎搞干嘛。。
不是发展，是偷
自己去查查美国半导体内需市场规模，比中国小很多。。
估计你没参与过多少半导体行业的决策层。。
还是那句话，放弃全球最大市场，对美国的半导体行业来说是自杀。。全美半导体行业，现在都在反对川普的瞎搞。。
挖人偷技术？别做梦了。。
技术防窃是门专业。挖人偷技术行不通了。还可能是个坑。
说个故事，当年苏27引进以后发现根据图纸做的东西不合格，问俄方总工他们也不知道，其实俄方总厂图纸就是错的。
后来俄方总部来人拿出正确的尺寸，但关照不许记录，不许修改图纸，只有做事的人才知道。
如果当年俄方总工叛变，就是带来全部图纸也没有用，做不出来的。总工很多事情不知道，不需要知道的不会知道的。
🛋️ 沙发板凳
成功走上了韩国的路，让韩国无路可走，欧耶！
欧耶好厉害！
预计下个月文在寅就会哭喊着加入中国成为吉林省一个县。韩国女团纷纷以能嫁给秒薪七毛的汉族猛男为荣，欧耶！
这个搞好了, 也动了美国的奶酪了. 美光, INTEL
你如果前后读一下EETIMES 新闻, 这个已经和所有的主流NAND 主控芯片厂家一起连调了, 美国和台湾的大厂的反馈都很Positive, 量产其实已经开始了. 下面就是良率和成本的问题了
三星凭这NAND和DRAM和晶圆代工, 已经和INTEL基本平起平坐了, 和INTEL交替作为世界的半导体头名. 能走上韩国的路, 可不那么容易啊
2018年的时候韩国NAND, DRAM坐地起价, 中国手机和电脑厂家一个季度就多付了30,40亿美元的成本, 你还没有任何讨价还价的能了.
现在DNAN降价是因为长江存储要出来了, 三星做不住了. 三星在西安投了150亿美元开厂做NAND, 不是没有原因的, 这个可是韩国以外的唯一厂子.
wangbin是谁? 这个是我的业余爱好, 国内有很多同学在做EE和IT.
NAND技术有啥难度吗？关键是密度和工艺的成品率。。。你确定你同学是学这个的？
3D NAND没有难度?
呵呵，果然是忘了抬头看id
我猜lz是某个id的马甲，发帖风格像
我觉得是wangbing
我没有马甲, 但是你信吗?
七毛一个帖子，收入不菲啊
我信你没有马甲，因为觉得你就是马甲。你的发帖和回帖风格真的是似曾相识哈
自干五, 没有钱的. 还得会英语, 能读懂专业文章, 难啊
VPN的钱是不是也得自己出？
哇，平均每天50多，一个7毛的话，日入35元人民币。恭喜楼主，你发财了！
是啊, 从Spectrum 登录的, 现在还在苦哈哈的做自己的网站呢. 再玩一会就睡了.
大家一起发财, 有了你们我才不寂寞. 调程序又是很累的, 多亏了你们陪我
所以是在美国的self-fuck-five咯？那就更令人作呕了。。。最看不起这类人。
说脏话了吧, 不NICE啊.
没看出来他说了什么脏话啊，再怎么都没你一个自干五脏不是么
语气是夸张了店, 但是进步还是很显然的.
我今天心情好, 不想和你然
你不是说你是马公么？怎么，硬件也是你的专业？非专业的也能懂夸张不夸张，进步不进步？
就一看新闻说话的门外汉，你跟他叫什么劲。。。
EE 出身的码农, 不是很正常吗? 你是学什么功的?
本论坛全能型人才多得很，[tuzki14]
所以说并不是业内人士咯，再说EE范围广了去了，连photonics都算在EE里
我什么时候说我是业内人士了? 半导体是是我的业余爱好, 你要不想看的可以不进来吗, 这个坛子一天上千的新帖呢.
ASML的大股东, INTEL, TSMC 和 SUMSUNG都是, 这个联盟暂时没有人能撼动.
所以我前面说硬件上你不是专业的有问题吗？
你要不喜欢我的回帖也可以不回复啊，这帖子里骂你的好几十条呢。
我不光在这里发帖啊, Reddit 也很好啊.
骂吧, 如果你的心里能好受点.
这个互掐很正常, 高端代工就这两家, 斗的你死我活, 但是他们和ASML的联系是牢固的.
3D NAND现在不需要EUV, 1x工艺即可.
(這些都是public knowledge, 沒有任何的insider information)
長江的NAND 是真的有他的技術和可行性。全世界四大NAND廠就美光不一樣。美光是 Peri-Under-The Array，省了~20%的晶圓面積。 今天先不管她ONO vs Charge Trap的不同，做peri under the array 是有代價的。代價就是 在做array 時，Peri Transistor 會持續受到上層製成的 Thermal Budget。這是美光和intel的FG NAND。他們有他們克服的方法，大叔不方便透漏。
可參考考古文獻:
https://marketrealist.com/2016/03/behind-microns-unique-architecture-3d-nand/
X-stacking 的原理是採用wafer bonding 來分開peri transistor 和 array。分兩片做。有效的避免掉thermal budget又可以享受20% 晶圓面積。可是這也是有代價的。分兩片做的cost 就高很多了。還有bonding 時的yield
這算是CMOS-Over-The-Array 吧
可參考考古文獻:
https://www.techinsights.com/blog/techinsights-memory-technology-update-iedm18
現階段cost 是一個問題。但是至少沒有侵權的麻煩。 長期我還是看好長江的發展前途的。
您是内行， 我这是一个EE出身业余爱好者，长江存储的技术能绕过大厂的专利池， 这个就不容易。 至于COST和良率， 我相信只要持之以恒的投入， 应该是能解决的。钱应该不是问题。 长江和长鑫如果都能做起来，还是很让人振奋的。
估计就是忽悠我这种不懂的
半导体 是你的业余爱好？笑死我了，你平时怎么实践你这个爱好的？我猜，也只能看看半导体新闻吧？
他们对海华的任务究竟是什么，要达到什么目的？挺纳闷的。
EXACTLY!
Behind the launch of any new product from China, it is a bunch of intelligent and industrious Chinese people.
What do those self-centered attackers know about it? They are just as myopia as Mr. T.
大家反驳的好激动...还好楼里有几个懂的。不管lz是什么原因发帖，长江存储这个难得是真提出了很大胆的跟其他几家不一样的方法，这个技术业界观望了好久，3dNand往高处做是趋势但太多技术难点，三星自己也不容易，更不用说micron谁的了。长江这个前阵子听说是结果还不错，希望能行。
你被分配的职业是调程序的？我觉得你去当演员不止赚五毛
哦呦，你们培训越来越厉害了
跟我学编程吧. MONEY大大的有. 一码在手天下走.
这个是国内芯片业的最大利好
来了几天就被安上五毛的帽子了. 五毛这个头衔不值钱啊. 以后改叫我们亿万富翁好不好, 那样听着多美
据我所知, 华为的智慧屏, 和手表已经都改成鸿蒙了,这个是微内核的, 比较容易定制化. 手机吗, 应该还是ANDRIOD.
你这因果搞反了，正是中国要发展这些高端产业，美国才发动毛衣战的。
没什么区别, 上层的UI和ANDRIOD是兼容的. 你用着没区别. 但是底层是微内核的, 这个和ANDRIOD是本质的区别
哦I也
哦耶
这毛病得治
效果根本不重要，看看windows cell phone的下场就知道，先占领市场那个通吃。
别跟这种外行科普好吗，公司有没有对敏感信息的培训
你原来不会的现在会了, 就是重大突破, 这个领域的玩家只有韩国, 美国和日本的3,4个厂家, 欧洲都一个没有, 这个每年上千亿美元的大市场, 比你家的铁锅有用多了.
川普一断货中国就从保险柜里拿出替代品，你信么？华为的高端PA，FPGA 根本无法找到替代品。当然它可以用质量差的小公司产品，问题是用次等元件怎么和其他公司竞争呢？新出的P40不是乖乖的用回美国PA芯片。至于5G设备需要的FPGA，美国是无论如何不会解禁的。没了高质量的元件，在国内还可以有政府强行帮华为开路，在海外市场呢？更别提EDA软件了，华为能一直用盗版吗？能一直不用EDA公司的支持吗？这些软件可是错误百出，需要无数support。目前看，中国压根没兑现采购协定，疫情过后，贸易战估计要升级，华为的台积电芯片来源再被一卡，哪来的生存空间？
FPGA这个国内已经有3,4个厂家可以做了, 只是局限在中低端而已. FPGA在基站里不是一定必须的. 华为和爱立信的里面全改成ASIC了, 成本比NOKIA低多了. Nokia的基站就用的FPGA, 现在傻了吧.
华为是一家公司, 你让他神魔都做? 拜托, 讨论要理性.
贸易战让美国认识到，绝不能让中共掌握供应链上的任何重要环节。如果中国控制了台积电和美国5G，那就不是宁可饿死猪也不进口美国大豆了，而是不出口芯片给美国或是让美国彻底断网了。
PA国内华为已经和三安合作了, 这个季度已经小量生产了, 等品控和成本控制之后, 不一定要用美国那两个厂家的了.