Elon Musk has become the focal point of US-China relations once again. The Wall Street Journal reported a couple of weeks ago that Tesla’s will be banned from military and some state government employees in China for fear of spying. Elon responded one day later at the 2021 China Development Forum that Tesla’s will never be used for spying.

As is often the case with Elon, there’s a mismatch between what he spent most of his time saying and what ends up getting reported. This happened last year when he appeared at the 2020 World Artificial Intelligence Conference in Shanghai for a Q&A session. The headlines that were produced from that event were all about his prediction that Tesla is close to achieving L5 self-driving capabilities. But what he spent most of his time talking about was how much “original engineering” Tesla plans to do in China and how aggressively he was hiring top-tier Chinese AI engineers -- a 15-minute recruiting pitch that has big implications for the on-going US-China technology competition. I discussed those implications in-depth last year in “Tesla, China, the ‘Tech Cold-War’”, so I won’t repeat myself here.

So what did we miss from Elon’s Q&A this time?

Lost in the Physics

The discussion was between Elon, Lan Xue (Dean of the Schwarzman College at Tsinghua University), and Qikun Xue (President of Southern University of Science and Technology. The two academics and Elon all studied physics, so most of the Q&A was a hardcord “nerd fest” about different types of renewable energy, difficulty in conducting academic exchange during today’s US-China tension, and how to make physics fun and cool in school. All three want more young people to become hardcore physicists to work on sustainable energy sources or building rockets, not computer programmers or financial engineers. One interesting tidbit was Elon wanting more students to study physics in conjunction with economics, so they could better grasp the real-life costs and tradeoffs of applied physics, like building electric vehicles or sending people to Mars.

The best video of the entire session is this one on BilliBilli. Elon’s words were dubbed with simultaneous translation into Chinese, because the audience is all Chinese.

The topic of Tesla possibly being used for spying was never brought up directly, but only tangentially in the last question, when Lan Xue asked Elon for his thoughts on the trend where every piece of technology can be treated as capable of “dual use” -- as in suitable for both civilian and military purposes.

Elon may not have the best dance moves, but he is quite adept at public relations. He took the chance to clarify that Tesla does not spy, and it’s in the company’s commercial interest to protect users’ privacy. He also shared his opinion on the pointlessness of the US TikTok ban (that wasn’t implemented) -- a view he has shared publicly before, so it wasn’t news.

Elon’s answer for this last question was the portion that received wide reporting, but the question’s context on “dual use” was mostly missing.

To be fair, I don’t blame mainstream media reporters for not writing about the future of physics education or how to replace fossil fuels with hydrogen. Those topics are more suitable for niche trade publications and blogs, and would put most people to sleep.

But the big story here is how regulators in the US, China, and elsewhere are all struggling to understand what “dual use” means in the 21st century and how to make policies accordingly. This struggle extends to most commercial companies, who have to navigate a fluid and increasingly incoherent regulatory landscape, while still trying to operate and sell products.

Putting it another way, do policy makers these days know enough about technology to carve out “narrowly tailored” restrictions, and not succumb to the mere possibility and fear of “dual use”, as well as the political pressure that comes with it?

(Note: I’m using the term “narrowly tailored” in a more plain language sense, not as it is used in a constitutional law sense.)

Narrowly Tailored

When it comes to cross-border tech products -- your Tesla’s, TikTok’s, iPhone’s, and Huawei’s -- the rationale of “national security” often gets over-applied and abused.

The TikTok ban is one such example. The ban would have been more reasonable and “narrowly tailored”, if the app was banned on all government-issued smartphones and other devices. But the ban was broadly construed to include all American users to protect them against spying and data gathering by a Chinese-made app. It was a fear-based ban, not an evidence-based ban. (You can read my multiple analysis on this issue here from last year, when the TikTok ban was in the headlines almost every day.)

By comparison, China’s ban against Tesla is more “narrowly tailored” to restrict use near military sites or by employees of certain sensitive agencies. The directive was not broadly applied to all government departments, let alone the entire population. It still made Tesla look bad, so Elon took the first opportunity he got to defend his company. Such reputational damage always affects a product’s trust with its civilian consumers, even if the ban is strictly for government. But as far as a piece of policy is concerned, the “Tesla ban” is more reasonable than last year’s “TikTok ban”. Military and sensitive government agencies have always received a higher level of protection for legitimate national security interests, no matter which country we are talking about.

This isn’t to say the Chinese regulators are some kind of model to follow, while American regulators are categorically bad at their job. In 2018, the US Congress passed a prohibition on federal agencies and government contractors from buying equipment made by several Chinese firms, like Huawei and ZTE, as a component of the Defense Authorization Act. Is this ban too broad? That’s a valid question to ask, but the provisions are, at least, somewhat “narrowly tailored”.

For most companies, helping one country spy on other countries is simply bad for business, unless the government is your business. This dynamic does not just apply in a cross-border context, but domestically as well.

There’s a common misconception that the Beijing central government is omnipresent and omnipotent. Whatever it wants, all layers of government and 1.4 billion people march in lockstep forward to deliver, like a military parade.

Simply. Not. True.

I illustrated the messy, call-and-response relationship between the central and municipal/district government in “China's ‘Semiconductor Theranos’: HSMC”. An even more complex web of incentives exists when dealing with domestic data collection.

Last April, the Financial Times published a detailed report on this complexity, which I wish more people read when they discuss China. The report illustrated many places of tension related to user data collection -- between companies like Alibaba and Tencent and provincial governments, between state-owned telecoms like China Mobile and local governments, and even between different departments inside the government, many of which do not trust each other enough to share data.

The popular narrative that whatever user data the Chinese government wants, it will get from Chinese companies does not reflect reality. As one Chinese tech executive shared in the FT report:

“We spend a lot of effort resisting Chinese authorities’ attempts to convince us to give over our data,”

Handing over data to a government, any government, is detrimental to a company’s commercial interests. This risk is especially high for publicly-listed ones, like Tesla, Alibaba, Tencent, Apple, Zoom, to name just a few. It also damages huge unicorns like ByteDance, whose public market debut is simply a matter of time.

On the other hand, there is less risk and accountability when you are not a publicly-traded company. That’s why Huawei -- an opaque, family-run business -- gets the brunt of the scrutiny. (That's also why I advocated for Huawei to go public in the US, if it truly wants to be seen as a trustworthy company.)

Elon, Earth’s Chief Recruiter

After Elon attempted to protect Tesla’s commercial interest at the China Development Forum and in a subsequent interview on CCTV, the Washington Post’s editorial board panned him for being “craven” to the Chinese government. I’m not sure how useful such a critique is.

So far, there’s no evidence that Tesla or any of Elon’s other companies have sold anything to aid and abet what’s happening in Hong Kong or Xinjiang. Panning every American executive, regardless of what their products or services do, for not bringing up Hong Kong or Xinjiang every time they do anything public with China seems too broad, and not so “narrowly tailored”. That is the job of senior diplomats and other government officials. Secretary Blinken is delivering on that.

What people tend to forget about Elon is his unique personal background. In some sense, he is a classic “global citizen”, having grown up and gotten bullied as a nerdy kid in South Africa, then briefly attending Queen’s University in Canada before transferring to UPenn. He moved to California, because he got into Stanford’s Physics PhD program and promptly dropped out to start a company and took full advantage of the first internet boom. When California was good to him, he stayed. When it wasn’t, he left to move to Texas.

Elon may be an idealist, but he is no purist. He is an ambitious pragmatist. He is not a national or geographical loyalist. Elon wants (and needs) all the countries on planet earth to get along, so he can live his dream of dying on Mars. He could care less about the ideological brinkmanship between the US and China.

What he cares about is getting more physicists, engineers, robotics programmers, and other technical talent trained to build EVs and rockets. And he will recruit from wherever he sees the potential.

Right now, America is still good to him, so he is still here. If and when that is no longer the case, he will leave, just like how he left California for Texas. He already has a Chinese “green card”.

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, Clubhouse (@kevinsxu).



每次埃隆出境,他花绝大部分时间说的话和最终被报道的内容经常不匹配。去年他在参与2020年世界人工智能大会之后就是这样,那次活动的头条都是关于他预测特斯拉即将实现L5自动驾驶能力的消息,但问答的绝大部分时间是埃隆谈特斯拉准备在中国做多少 "原创工程" (original engineering),所以在积极招聘中国顶级的人工智能工程师。整个过程其实是个15分钟的招聘广告,而且对整个中美科技竞争有很大影响。去年我在《特斯拉,中国,“科技冷战”》这篇文章中中也深入讨论了这些影响,就不在这里重复了。





特斯拉可能被用于间谍活动的话题并没有被直接提起过,直到最后一个问题时,薛澜问了埃隆他对每一项技术都可以被视为能够 "两用"(即适合民用和军用)这种趋势的看法。


埃隆对最后这个问题的回应是最终被广泛报道的部分,但问题背后对 "两用" 的大背景却缺失了。


但更值得报道的大问题其实是,美国、中国和其他国家的所有监管机构都无法很好地去理解 "两用" 在21世纪科技里到底意味着什么,又如何制定相应的政策。这种挣扎延伸到大多数企业里,它们必须在多变的、越来越不一致的监管环境中努力适应,同时保持继续运营,销售产品。

换个方法说,当今的政策制定者是否对科技有足够的了解,能够设计出一套 "狭义定制"(narrowly tailored)的限制,而不屈服于仅仅是可能的 "两用" 所带来的恐惧,以及随之而来的政治压力?

(注:我在本文中使用的 "狭义定制" 一词是从通俗语言层面上的理解,而不是解析美国宪法层面的使用)。


说到跨境科技产品 —— 大家熟悉的特斯拉、抖音、iPhone和华为 —— "国家安全" 这个理由往往被过度使用和滥用。


相比之下,中国对特斯拉的禁令要 "狭义定制"得多,仅限制在军事地点附近或某些敏感机构的员工。该指令并没有传达到所有政府部门,更没有包括全国民众。当然,还是给特斯拉品牌形象画了个大污点,所以埃隆第一时间抓住机会为自己的公司辩护。这样的名誉损失总是会影响到产品在大众消费者心中的信任度,即便禁令只针对政府。但从评估一项政策而言,"特斯拉禁令" 比去年的 "抖音禁令" 要更加合理。不管是哪个国家,军事和某些敏感政府机构出于合理的国家安全利益,总会受到更高级别的保护。

这并不是说中国监管机构是个值得效仿的模范,而美国监管机构则断然无能。2018年,美国国会通过了一项禁令,禁止联邦机构和政府承包商购买华为、中兴等几家中国企业生产的设备,当时是以《国防授权法》的一部分通过的。这个禁令是否太宽,是个可以探讨的问题,但起码在某种程度上是 "狭义定制" 的。




我在 《中国的 "半导体Theranos":弘芯》一文中,仔细形容了中央与市和区政府之间常常混乱的、"号召+响应"关系。在收集用户数据这个问题上,政府于企业的关系则更为复杂。

去年4月,英国的《金融时报》发表了一篇关于这个复杂性的详细报道,我希望更多人在讨论中国时能先看看这篇报道。该报告形容了许多在用户数据收集方面多层的对峙对立关系 —— 像阿里和腾讯等公司与省政府的关系,像中国移动这种国有电信与地方政府的关系,甚至是不同政府部门之间的关系:许多部门相互之间并不信任,不愿分享数据。






在埃隆试图在中国发展论坛上以及随后接受的央视采访中,保护特斯拉的商业利益,但《华盛顿邮报》编辑部抨击了他在中国政府面前的 "懦弱"。我不知道这种批评有多大用处。

目前没有证据表明特斯拉或埃隆的其他公司出售了任何产品用于协助在香港或新疆发生的事情。抨击每位美国高管,不管他们的产品或服务是做什么的,只因为他们每次在中国公开做任何事情时,不提香港或新疆,这似乎太宽了,并没有任何 “狭义定制”。这是外交官和其他政府高管该做的工作,布林肯外交部长已经在这一点上兑现了

人们往往忘记埃隆独特的个人背景。从某种意义上说,他是一个典型的 "全球公民",从小在南非长大,是个经常被其他孩子欺负的书呆子,然后在加拿大的皇后大学短暂就读,之后转到美国的UPenn。他搬到加州,是因为考上了斯坦福的物理系读博士,随后又退学创业,在互联网热潮的第一波赚了一把。当加州对他好的时候,他就留下。当情况变了之后,他就立即离开了,搬到了德州。



现在,美国对他还是不错的,所以他还在这里。如果有一天情况变了,他就会离开,就像他离开加州去德州一样。他已经有中国 "绿卡" 了呢。