This issue covers the period between August 31 - September 6, 2020 with six news stories – three from English language sources, three from Chinese language sources. Disclaimer: all translated article titles are done by me, not official translations from the media outlets.

Before you go on, please check out last week's deep dive post: "What is TikTok Worth to Whom and Why?"


Trump administration weighs blacklisting China's chipmaker SMIC” (English Source: Reuters)

My Thoughts: if SMIC is officially blacklisted, it’ll be a death knell to China’s fledgling semiconductor sector in the short-term, in much the same way that the previous three rounds of sanctions have been a death knell to Huawei. SMIC, the “national champion” of semiconductor manufacturing, has already faced “unofficial sanctions”, like the Trump administration’s lobbying of the Dutch government to not allow a sale of lithography equipment (for chip manufacturing) by ASML to SMIC. Meanwhile, HSMC, another Chinese chip foundry that has attracted a ton of investment and government subsidies, is not pulling its weight (more below).

India Bans 118 Chinese Apps as Indian Soldier Is Killed on Disputed Border” (English Source: New York Times)

My Thoughts: India’s latest round of Chinese app ban made the news in many outlets last week. The most headline-grabbing ban among the apps was the popular battle-royale style game, PlayerUnknown’s Battlegrounds (PUBG), whose South Korean maker has close ties to Tencent. But even more benign apps, like one that scans business cards, were included in this ban. Coupled with ongoing border tension, India is basically off limits as an expansion and investment destination for Chinese tech companies in the short-term. The animosity appears both serious and shrill. To get a feel of the emotion, I recommend you watch a short clip of Indian cable TV news I included (above) talking about this ban. Indian cable TV is known for its theatrical debates, drama, and anxiety-inducing headline animations – CNN and FOX looks quiet and tame by comparison.

China to Plan Sweeping Support for Chip Sector to Counter Trump” (English Source: Bloomberg)

My Thoughts: While we won’t know all the details of the 14th five-year plan until later this year, $1.4 trillion dollars look to be the price tag China will commit to its domestic chip sector in the next five years, in order to establish technological self-sufficiency. The focus will be on the third generation of semiconductors -- chips that go into 5G equipment, electric and/or self-driving vehicles, and military equipment. China has been importing $300 billion worth of semiconductors for the last two years and is on pace to do the same this year (graphic below), though we don’t know if all the US sanctions will artificially lower that amount. Regardless, the vulnerability of being so dependent on foreign technology is painfully obvious. The money will clearly be there, but will the people deliver?

Doubtful Cloud Shrouds HSMC: Constructions Halted, More than 500 Employees Still Working” (Chinese Source: First Finance)

My thoughts: If you read the previous section before this one (like I hope you would), then you will notice that this article is a direct foil to the Bloomberg one. Reporters on the ground have been digging into what’s happening at HSMC (Wuhan Hongxin Semiconductor Manufacturing), and things don’t look so...productive. HSMC, one of the more high-profiled chip foundry upstarts, has attracted 128 billion RMB (roughly $18 billion USD) of investment and subsidies since its founding in 2017 and hired a legend in the semiconductor space, Chiang Shang-Yi, to be its CEO. (Chang was TSMC’s first CTO, a topic I explored in-depth in “RISC-V, China, Nightingales”.) Yet, its factory location in Wuhan has not been worked on for eight months due to payment delays to the contractors. Meanwhile, a makeshift office was put up 500 meters from the construction site (see picture below), where employees continue to “do work”. This reporting further revealed that the majority owner of HSMC is a Beijing-based company, whose founders and team have no experience building a chip foundry (or anything of scale). Yet, the same company also spun up another chip maker in the city of Jinan in 2018, called Quanxin Integrated Circuit Manufacturing (QXIC). Both HSMC and QXIC have been aggressively luring TSMC engineers with 2 to 2.5-times their current salary, which was reported by the Nikkei Asian Review and discussed in a previous Interconnected Weekly issue. Besides hiring and roping in more investments and government subsidies, neither of these two foundries have done much chip-making. Is this what China’s semiconductor future looks like?

HSMC's makeshift office

Did we really have a late start?” (Chinese Source: Quantum School)

My thoughts: This commentary by Quantum School, a WeChat public account focused on the natural sciences, is a critical self-reflection of why China is lagging so far behind on semiconductor technology and infrastructure software, like operating systems. A common excuse shared in the Chinese tech community is that “we [China] had a late start”, which the author thinks is total BS. According to this piece, China began planning to develop a semiconductor sector in 1956 (two years before Japan did) and started building lithography equipment in 1965. During this current wave of massive investment (and nationwide insecurity), HSMC used a hefty amount of government subsidy to purchase top-tier lithography equipment from ASML to publicly show off its conviction, but later loaned off the same equipment to pay off debt. It’s difficult for me to independently verify all the claims made in this post. But for what it’s worth, this critique was widely shared and read, having reached the “100k+ Reads” threshold on WeChat within two days of publication.

Absorbing TikTok will not be easy: Hard to sell core technology and team” (Chinese Source: InfoQ)

My thoughts: This article makes a good complement to my deep-dive post last week What is TikTok Worth to Whom and Why? With China updating its own “entity list” to include technologies that most certainly exist in TikTok, whoever buys TikTok will likely not get the app’s AI algorithms and recommendation engine. Even though having the users and data is still valuable (arguably more valuable than in the algorithm in my opinion), operating TikTok in this “hybrid” fashion is practically impossible. To make the app work (before the buyer builds a new recommendation engine), American user data will have to be transferred to China, so the algorithms can be “trained” continuously and generate recommended videos to serve up, which defeats the whole purpose of protecting the American people’s data from Chinese access. At this moment, I would not be surprised if the whole deal falls apart and TikTok just gets banned outright.

If you like what you've read, please SUBSCRIBE to the Interconnected email list. To read all previous posts, please check out the Archive section. New content will be delivered to your inbox (twice per week). Follow and interact with me on: Twitter, LinkedIn.

本期《每周互联》总结概括的时间段是:2020年8月25至9月6日,包括本作者挑选的六条新闻:三条原文是英文,三条原文是中文。声明:所有翻译的文章标题都是我做的翻译,不是官方翻译。

还希望大家有空看看我上周最新的深度分析文章:《TikTok为什么,又对谁,有什么价值?》


特朗普政府考虑将中芯国际列入黑名单”(英文来源: 路透社)

我的想法:如果中芯国际被正式列入黑名单,短期内将对中国刚刚起步的半导体行业敲响丧钟,这与此前的三轮制裁对华为的影响很相似。做为半导体制造业里的“国家队主力”,中芯已经经历过“非官方制裁”,比如特朗普政府游说荷兰政府不允许ASML向中芯出售用于芯片制造的光刻设备。与此同时,另一家吸引了大量投资和政府补贴的芯片铸造厂,武汉弘芯,好像还什么都没干(详见下文)。

印度封杀118个中国app,因为印度士兵在有争议的边界被杀”(英文来源: 纽约时报)

我的想法: 印度最新一轮对中国app的禁令上周在多家媒体都有被报道。在被禁的app中,头条点击价值最高的游戏,PUBG,因为此游戏的韩国制造商与腾讯关系密切。但即便更 “平淡日常” 的一些应用程序,如一个扫描名片的app,也被包括在这项禁令中。此禁令加持续的边境紧张局势,使印度在短期内基本上不会看到更多的中国科技企业在那里扩张和投资。印度对中国的敌意既严肃又尖锐。如果想感受一下这种情绪,我建议您看一段印度有线电视报道此禁令的新闻片段(上面)。印度有线电视以富有戏剧化的“导论”和不断令人焦虑的头条动画为名,相比下CNN和FOX要“平静”多了。

中国计划全面支持芯片行业以对抗特朗普”(英文来源: Bloomberg)

我的想法:  虽然要到今年晚些时候才能知道第14个五年计划的所有细节,但国家为了实现科技“自主可控”,而承诺在今后五年投入1.4万亿美元于国内芯片行业,看似差不多敲定了。重点将放在第三代半导体上——用于5G设备、电动或无人驾驶车以及军用装备的芯片。过去两年,中国一直在进口价值3000亿美元的半导体,今年应该也会(下图),不过还不知道美国的制裁是否会降低这一数额。无论如何,如此依赖外国技术的脆弱性是显而易见的。钱应该是有的,但人会兑现出成果吗?

千亿弘芯爆雷疑云:土建工程停摆,500员工照常工作”  (中文来源: 第一财经)

我的想法: 如果您已经读了上一段想法(像我希望的那样),那您会注意到这篇文章是Bloomberg那篇的直接陪衬。记者到武汉弘芯的工地现场做了深入调查,发现公司的进度基本不存在。作为一家小有名气的新兴芯片铸造厂,弘芯自2017年成立以来,已吸引了1280亿元人民币(约180亿美元)的投资和补贴,并聘请到了半导体领域的传奇人物蒋尚义担任CEO。(蒋是台积电的第一任CTO,我在《RISC-V,中国,夜莺》中深入探讨了这个话题。)然而,由于给承包商的付款延迟,其位于武汉的工地已经八个月没有开工。同时在距施工地500米的地方搭建了一间临时办公室(看以下照片),员工们继续在那里“工作”。这条报道进一步透露,弘芯的大股东是一家北京的科技有限公司,其创始人和团队没有建造芯片铸造厂(或任何有规模的工厂)的经验。然而,2018年,这同一家公司还在济南成立了另一家芯片制造厂,济南泉芯(QXIC)。武汉弘芯和济南泉芯一直在以2到2.5倍的薪水努力诱惑台积电工程师,这在《日经亚洲评论》上报道过,我也在前一期的《每周互联》里讨论过。除了雇人和圈投资和政府补贴,这两家芯片铸造厂看似都还没有在造多少芯片。中国半导体的未来难道就张这个样子吗?

武汉弘芯临时办公室

我们起步真的晚吗?”  (中文来源: 量子学派)

我的想法: 量子学派是一个关注自然科学的微信公众号,它的这篇评论是对中国为什么在半导体技术和基础设施软件(如操作系统)上如此落后的一个严厉的自我反省。在当前的中国科技界,一个常见的借口是“我们(中国)起步晚”,但作者认为这个借口违背事实,都是推卸责任的废话。根据这篇文章,中国从1956年开始就计划发展半导体行业(比日本早两年),并于1965年开始制造光刻设备。在当前这新一波大规模投资(以及全国上下的不安定和不自信)中,弘芯利用巨额政府补贴从ASML购买了顶级的光刻设备,以公开炫耀自己“报国”的壮志,但后来却把设备借出去还债。我很难独立核实这篇文章中的所有说法。但值得一提的是,这篇评论已被广泛分享和阅读,发表后两天内就超过了微信的“10万以上阅读量”。

吃下TikTok并不容易:技术和工程师团队均难以出售” (中文来源: InfoQ)


我的想法:这篇文章很好地补充了我上周的深入分析文章TikTok为什么,又对谁,有什么价值?随着中国监管部门对自己“实体清单”的更新,包括了TikTok产品里肯定有的一些核心技术,无论谁最终买到TikTok,都很可能得不到TikTok里的人工智能算法和推荐引擎。尽管拥有用户和数据仍然很有价值(我甚至认为比算法更有价值),但以这种“混合”方式操作TikTok实际上是不可行的。为了让这个app正常运作(起码在买家打造一套新的推荐引擎之前),美国用户的数据必须被传输到中国,这样AI算法和模型才能不断受“训练”,同时继续生成推荐视频来服务用户,这就违背了保护美国人民的数据不受中国访问的最初目的。客观的看目前的状况,如果卖TikTok的所有交易都泡汤,app被彻底严禁,我不会感到惊讶。

如果您喜欢所读的内容,请用email订阅加入“互联”。要想看以前所有的文章,请看《互联档案》。新文章将会直接送到您的收箱(每周两次)。请在TwitterLinkedIn上给个follow,跟我互动交流!