This issue covers the period between September 21 - 27, 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: "300 Years: Huawei's Open Source Strategy"

Jim Wilson/The New York Times

Facebook Takes Down Fake Pages Created in China Aimed at Influencing U.S. Election” (English Source: New York Times)

My Thoughts: The title of this article probably oversold China’s effort and investment in influencing this year’s US election. Reading the article closely, it’s clear the Pages that Facebook took down have had little success in gaining an audience and were made by groups that may not even be affiliated or funded by the government. The topics these Pages chose are mostly to plant division among the American electorate, not explicitly supporting or undermining the presidential candidates -- a strategy that the Russian government applied effectively in 2016 and is applying again in 2020. So as far as this election is concerned, worry of meddling from China is a valid but relatively minor one. If the goal is to sow division among Americans, the American president is doing a better job of that than just about any foreign power.

Microsoft teams up with OpenAI to exclusively license GPT-3 language model” (English Source: Microsoft Official Blog)

My Thoughts: This licensing agreement between Microsoft and OpenAI is somewhat expected, given that Microsoft is a big strategic investor in OpenAI and all of OpenAI’s new models and APIs are already running on Azure. It’s still noteworthy because I think it signals a tectonic shift in the cloud industry landscape, even though it'll take a few more years to materialize in earning statements. Because the market leader AWS had a 7-year-long head start, every other cloud platform is trying to both catch up in terms of market share and differentiate in terms of product. This exclusive agreement will give Microsoft Azure the differentiation it needs to become the “AI Cloud” in the market. AI has always been the calling card of Google, but Microsoft appears to be moving much more quickly to productize AI into its cloud. For a deeper dive on what I think of GPT-3 the technology, please see “GPT-3: An Abstraction of Time vs Money”.

Merkel Resists Full Ban on Huawei, Making Germany an Outlier” (English Source: Bloomberg)

My Thoughts: Merkel’s unwillingness to “follow the herd” in banning Huawei is a welcoming sign, not necessarily for Huawei, but for the larger challenge of regulating cross-border technology product overall. Germany’s regulatory framework is still being deliberated, which does not mean Huawei won’t get banned. What it does mean is that Merkel is intent on doing the hard work of designing a generalized framework beyond singling out one company. It’s work that major countries, including the US, China, and UK have thus far all failed to do, while other countries look for guidance and leadership. Just like software, if a regulatory framework is well-designed and well-reasoned, it can be exported too!

As Apple quietly cuts back on manufacturing in China, what will happen next?” (Chinese Source: Geek Park)

My thoughts: While it’s hard to verify the information in this post, it looks like some of Apple’s lower end supply chain is indeed moving from China to India. Earlier this Spring, I wrote a post questioning the practicality of a supply chain decoupling by Apple away from China -- given both the tight integration and high manufacturing quality in China. What I failed to account for is just how complex the supply chain is that goes into an iPhone, thus certain less value-add portions could be peeled away to lower cost, less experienced countries. Interestingly, the manufacturers doing the moving are still the Foxconn’s of the world -- same companies expanding into new locations. Another interesting angle is the places they are expanding to -- not just India, but also Vietnam and Mexico. So “decoupling” may end up being “location diversification” for these manufacturers and suppliers to both reduce cost and strengthen their resilience against future geopolitical shocks.

HSMC's makeshift "factory"

HSMC suspected of ‘removing and restarting’; Why do domestic semiconductor manufacturers encounter so many difficulties?” (Chinese Source: InfoQ)

My thoughts: I’m mildly regretting giving HSMC (Wuhan Hongxin Semiconductor Manufacturing) so much ink in my first deep dive into the Chinese semiconductor space in “RISC-V, China, Nightingales”, because it now looks to be a total sham. This article shared that HSMC is supposedly sending its top engineers (any one who hasn’t quit yet) to the province of Guangdong under a different semiconductor company, while planning to shut down its operation in Wuhan. Guangdong, being the entrepreneurial and enterprising province that it is, has seen a massive spike in new semiconductor companies in the last eight months, along with quite a few other provinces. Will this “gold rush” for government subsidies doom China’s search for technological self-sufficiency?

Who is OpenHarmony for?” (Chinese Source: YuanChuan Tech Forum)

My thoughts: This commentary is even more pessimistic than my own thoughts on Huawei’s OpenHarmony and overall open source strategy discussed in last week’s deep dive post: “300 Years: Huawei's Open Source Strategy”. The pessimism is not without precedence. The mobile operating system landscape is littered with failures to unseat the duopoly of iOS and Android, like Microsoft’s Windows phone and Samsung’s Tizen. This post has the correct analytical framework in that a mobile operating system can only take off if it has a strong hardware ecosystem to lean on, which Huawei is lacking due to all the semiconductor sanctions. Carving out a niche in IoT or AIoT (AI + IoT) may help OpenHarmony’s adoption somewhat, since it was originally built to be an IoT operating system. IoT technology, both software and hardware, has a relatively lower barrier to entry. Huawei is in this for the long haul though, so no sentiment, optimistic or pessimistic, will be proven out for years or decades, if not centuries.

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.



Jim Wilson/The New York Times

Facebook删除来源中国的假网页,旨在影响美国大选”(英文来源: 纽约时报)


微软与OpenAI合作独家授权GPT-3语言模型”(英文来源:  微软官方博客)

我的想法: 考虑到微软是OpenAI的一个大战略投资者,而且OpenAI的所有新模型和API都已经在Azure上运行,微软和OpenAI的这项合作多多少少是在预料当中的。但仍然值得注意,因为我认为这标志着云计算行业格局的一个结构性大转变,尽管还需要几年时间才能在盈利报表中体现出来。由于整个市场的领头羊AWS有着长达7年的领先优势,其他所有云平台都在努力追赶,为了占据市场份额,还要达到产品差异化。这项独家合作协议将使微软Azure成为市场中的“AI云”,对差异化极有力。AI一直是谷歌的强项,但微软似乎更快的把AI在自己的云上产品化。关于我对GPT-3这门技术的更深入看法,请看《GPT-3:时间与金钱的抽象》这篇文章。

默克尔反对全面禁止华为,德国成为一个异类”(英文来源: Bloomberg)

我的想法:   默克尔不愿意“跟风”禁华为,是件好事情,但不一定是对华为好,而是对整个监管跨境科技产品这个更大的挑战好。德国的监管框架仍在审议中,这并不意味着华为不会被禁。更深层的意义是,默克尔愿意努力设计一个可以通用的监管框架,而不是把矛头指向某一家公司。这项工作是几个大国,包括美国、中国和英国迄今为止都未能做到的,而其他国家正在寻找导向。就像软件一样,如果一个监管框架设计得好、理由充分,它也是可以输出的!

当苹果正在悄悄减少中国制造,接下来会发生什么?”  (中文来源: 级客公园)

我的想法: 虽然很难核实这篇文章中的所有信息,但看起来苹果的一些低端供应链确实正在从中国转移到印度。今年春天早些时候,我曾写过一篇文章,质疑苹果与中国供应链脱钩的可行性,考虑到与中国生产业的紧密关系和高制造质量。我没有很好的理解的是,做一台iPhone背后的供应链有多复杂,因此某些附加值较低的部分是可以被移到成本更低、虽然经验较少的国家。有意思的是,进行迁移的制造商仍然是像富士康的这种大企业——同样的公司只是去了新的地点扩张。另一个有趣的是它们正在扩张的地方不仅仅是印度,还有越南和墨西哥。因此,“脱钩”可能最终成为这些制造商和供应商的”地点多样化”,既能降低成本,又能增强抵御未来地缘政治冲击的能力。


武汉弘芯疑似“换地重生”,国产造芯势力为何总遇困局?”  (中文来源: InfoQ)

我的想法: 在《RISC-V,中国,夜莺》中,也是我深入分析中国半导体领域的第一篇文章,写了不少关于武汉弘芯的内容。现在有点后悔,因为这个公司看似是个大骗局。这篇文章说,据推测,武汉弘芯将把自己的高级工程师(如果还没辞职的话)都派到另一家在广东的半导体公司,同关闭其在武汉的业务。广东,作为一个很擅长创业和做生意的省份,在过去的八个月里,冒出了一大批新半导体公司,许多其他省份也有类似现象。这种为了政府补贴的“淘金热”会不会给中国寻求科技独立的努力带来厄运?

鸿蒙给谁用?” (中文来源: 远川科技评论)

我的想法:  这篇评论比我对华为的鸿蒙OS和整个开源战略的看法还要悲观些(我的观点在上周的深度专栏文章《300年:华为的开源战略》中有详细讨论)。这个悲观并非没有根据和先例。在移动操作系统领域里,有过很多想挑战iOS和Android的双头垄断而惨败的例子,如微软的Windows 手机和三星的Tizen。这篇文章的一个分析角度是正确的,一款移动操作系统必须能依靠一个强大的硬件生态才能腾飞,而华为由于受到多种半导体有关的制裁而缺乏这一硬件生态。在IoT或AIoT(AI+IoT)中打开一套赛道可能有助于鸿蒙OS的推广,因为它最初就是为物联网设计的操作系统。物联网技术,无论是软件还是硬件,门槛相对较低。不过,华为在这方面的战略是很长期的,因此无论是乐观还是悲观,要看到现实的认证可能需要数年,数十年,甚至几百年。