preamble
I recently (months ago) attended a day-long series of panels at Johns Hopkins SAIS’ new building in D.C., entitled “The China Debate We’re Not Having: Politics, Technology, and the Road Ahead”. The title of this assemblage of discussions struck me as hilarious, given that those are the only things people in Washington are talking about. Why, in the past month alone, I attended a talk at Georgetown about chips, semiconductors, and Chinese industrial policy and another about the future of US Tech cooperation in East Asia.
Can we please talk about literally anything else? It doesn’t have to be an obscure topic. Tell me about the private sector. Do a deep dive on Chinese banking. On regulations and standardizations. Tell me about local politics, about the environment, about minority policy. All of these topics are getting headlines right now due to the 2026 Government Work Report (second source) and the new “Ethnic Unity and Progress Law” (CFR has a good backgrounder on it), but they aren’t getting any think tank air-time.
Tell me about how international companies are navigating data-sharing laws. Tell me about how China is navigating its relationships with Australia, Mongolia, Vietnam, Germany, Russia, India, etc. Tell me the daily workout regime in the PLA, tell me about China’s goals for small nuclear, tell me how STEM education in China is structured? Can we do a panel on Deng Xiaoping? Can we have a single panel that has members of the private sector as discussants? Can any think tank right now give me a rundown on Taiwanese politics? Who is worried about CBAM? Can somebody get them to stop beating me to death with critical minerals and the AI race? You can tell me about demographics, I don’t care! At this point, I’m even interested in knowing what the Chinese representative to the UN, Fu Cong, ate for breakfast!
One of my favorite moments was at the full-day Johns Hopkins event, where the most important panel of the day ended with the question of “Who do you think will replace Xi Jinping?” Every single panelist on stage had no answers and wouldn’t even offer their best guesses. The panelists went, “Xi Jinping will be the next president, ha-ha,” then, “Someone we don’t know yet,” then, “If I told you my guess, I’d be painting a target on their backs,” and then, “You focus too much on who is president instead of the policies.” Really folks? Are you kidding me? Just throw a name out there! It’s not that hard! You don’t have to be right, just not a coward. And if you don’t know, then just say so!
Am I going mad? I feel like every think tank and university is hosting its own event discussing China, the US, and tech, and the only thing unique about each event is the quality of the catering! (Also, massive shout out to the caterers at SAIS and Georgetown for providing excellent sandwiches and wraps—absolutely nothing to complain about there).
In honor of all the tech talks I attended this spring, I’ve synthesized a month of think tank info gleaned from these events, the key ideas being thrown around, and a list of topics to look up later:
ALLIES AND DETERRENCE
Everything in this section can be boiled down to a single bullet point:
- The US is being very silly with its allies right now. Not good. AAAAAAAAAAAAAAAAA!!!
To elaborate:
- There is deep hypocrisy in the US of how the export regimes are being handled. Export controls only work if allies work together. Under Trump, the US’ export control regime has dropped off in quality, but his administration still demands that allies maintain that high standard.
- Watch what happens with AUKUS. It’ll definitely be interesting. The intelligence communities of Australia, the UK, and the US are very interdependent, but the military-industrial complexes are not. (But they should be, to further enhance credible deterrence). In fact, there should be greater interoperability amongst allies, in areas beyond the military sector. Tech and industrial technology should be easily transferrable and useable between allied countries. Think “COCOCO” COdevelopment, COproduction, and COsustainment.
- How can the US work with allies to achieve economies of scale outside the USA?
- Who is Trump Bringing Together, and why is it Korea and Japan? When it comes to their relationship with the US, Korea and Japan come to the bargaining table without a lot of alternatives. In the face of slow growth in commercial sectors and the demographic decline, Korea and Japan see American markets as a safer alternative to China. Ironically though, Trump’s tomfuckery has in fact forged a tighter relationship between Korea and Japan, two countries who historically have had bad relations. It might be surface level, but it’s something.
BARGAINING CHIPS
This section talks a lot about Chinese industrial policy. For an overview of US industrial policy, you may want to peruse this beautiful explainer of the CHIPS and Science act by Victoria Cooper at Australia’s US Studies Centre. Now on to China:
By far the most focused discussion of chips, semiconductors, and industrial policy was Eric Harwit’s talk at Georgetown University. He works at the East-West Center and Hawaii University. If only so many people attended his talk as attended USCET + Young China Watchers’ discussion! Alas.
In 2014, China created the China Integrated Circuit Industry Investment Fund (aka ICF, aka ‘The Big Fund’, aka the 国家集成电路产业投资基金), which in 2024 received a third phase of funding. The first phase was a fund of 138.7bn yuan, the second phase was 204bn yuan, and this third and final phase was 344bn yuan. China clearly cares about having a domestic, militarily competitive semiconductor industry.
China has what I (and perhaps Dr. Harwit) would tentatively call a monopoly on legacy chips (chips of 28nm or larger). They produce more of these chips, which are used in everything from cars to household appliances, than the rest of the world combined. The chips that people generally freak out about are 4nm or smaller. Taiwan’s TSMC has the chokehold on 3nm and 2nm processes, and you should read more about them over on Wikipedia, honestly.
China’s R&D/industry-building ecosystem is proving very successful, but there is also enormous waste in the system. According to a 2022 report (which I am struggling to find), 2.3 billion dollars went to projects which never made a single chip. This data point is shrug-worthy—maybe the type of research they were doing didn’t require any chips to be manufactured, but I assume that China’s semiconductor industry is in the same position as its EV industry, where dozens of companies leapt on the EV subsidies, and now that the subsidies are going away, they’re dropping like flies because their product isn’t good enough to compete. All the R&D money that had gone into building those EVs had gone to waste. That’s what I assume the semiconductor industry will do after this final batch of funding runs out.
But Chinese industrial policy has certainly carved out some big wins in the form of companies like Huawei. Huawei makes 67% of its revenue domestically, 21% in the Europe/Middle East/African market, 6% in the Asia Pacific, and 5% in the Americas. In 2023, it spent 23% of its revenue on R&D. In addition to phones, Huawei is also big into chips and semiconductors. You could write a whole article about Huawei’s new Ascend 910B Chip, which is supposedly equivalent to Nvidia’s H100 chip, but I am too lazy to do that.
Especially for China’s cloud and AI companies, these are the kinds of chips that really matter. Currently Nvidia is still extremely dominant, but Harwit noted that within five years, experts predict that Nvidia’s market share in China will fall as indigenous Chinese companies catch up and/or steal chip technology from Taiwan’s chip ecosystem (which Nvidia is a part of). One fun fact I learned was that Nvidia is very good at keeping its employees from being lured away by potential rival companies. Nvidia has 4,000 employees in Mainland China, and of those, they have a 0.9% employee turnover rate. Due to export controls on the most advanced chips, Nvidia has been making ever-so-slightly-shittier chips for the Chinese market which are as good as they can possibly be without infringing upon the law.
Recently there has been a re-evaluation of export controls, with Howard Lutnick saying “The idea is the Chinese are more than capable of building their own. (…) You want to keep one step ahead of what they can build, so they keep buying our chips.” Even though Nvidia still has the go-ahead to sell some pretty good chips into the Chinese market, their share of revenue in the PRC has tumbled from 21% in Oct. 2023 to 12% in Oct. 2024. When Nvidia was cleared to sell H20 chips to the Chinese market, the government warned companies against using them for national security concerns.
While discussing Chinese reliance on American technology, Harwit mentioned China’s Directive 79. Also known as ‘Delete A’ (delete America), the directive calls for Chinese companies to escape American tech dependence by 2027. For some reason, the internet has varying opinions about this. WSJ was the one to break the story. In 2023, a version of the document was published, which made no mention of America specifically. Read it here. Of course, there could be multiple versions of the same document, one of which is deemed more suitable for the public. But anyway.
What is the rest of the world doing, seeing as there is huge geopolitical risk from having China or Taiwan control most of the semiconductor/chip manufacturing power? The US wants to start manufacturing chips domestically, but it (along with many other countries) is running into the issue that there just aren’t enough highly skilled technical workers to make that happen.
Japan might have a slight headstart on semiconductor manufacturing, because they still have advanced manufacturing expertise left over from the ‘80s and expertise in chemicals for semiconductor manufacturing and photoresist. In the ‘80s, Japan made up about half of the semiconductor industry, whereas now it’s dropped to 10%. However, any country trying to break into the semiconductor industry at this point will face the huge web of existing partnerships. Why buy from a new Japanese semis company when you have a good history working with a Taiwanese one?
TSMC is planning to build semiconductor factories in Germany and the USA, but there again, they run into the issue of finding those highly skilled workers.
One final company of note is ASML, a Dutch photolithography company who is the only supplier of the machines used to make the most advanced chips. The machines perform a process known as EUV Lithography (Extreme ultraviolet lithography). These machines are gigantic, and (so I assume) are harder to steal and smuggle. You can’t just pretend to be a Southeast Asian company, order a bunch of chips, then stuff them into a suitcase and hop on a flight to China (which is apparently how high-grade chips get past US export controls).
For even more about China and semiconductors, check out this report.
ECONOMIC COERCION
The talk (of which all evidence has been erased from the internet / their promotion of the event was lacking) titled The Future of US Technology Cooperation in East Asia was also not a waste of time, as they discussed economic coercion in addition to the usual suspects.
There is a volunteer taskforce with a shoestring budget within the US government that provided consulting services to nations looking to avoid Chinese economic coercion. For example, if Country A wants to say no to a deal with Chinese tech giant Huawei, that country might approach the US for advice. That consultancy analyzes what Chinese coercion is likely to hit and how to diversify and shift the supply chains in advance of any Chinese action being taken. Then, Country A is safe to say no to Huawei. The country may also be able to forge closer business relations with advanced democratic allies to make up for lost Chinese revenues. Learn more about the consultancy here, and the case that came up most frequently, Lithuania. Another example which came up was the case of Filipino banana exports to China. Bananas are one of the Philippines’ largest agricultural exports, and during a period of tension between them and China, the PRC stopped all imports of Filipino bananas on the grounds that they were infested with pests. In discussing the situation at the Hopkins event, experts implied that the US bought up the extra bananas to help mitigate the damage to the Filipino economy, but this article by these guys notes that diverting the flow of bananas from the Philippines to the US was not feasible as a long-term solution.
Starting in 2025 with the Trump Administration backing down in the critical minerals fight, there has been a shift among mid-size and small-size countries as they began to wonder if even the US fears Chinese economic coercion. This contributes to the sense that the US is becoming a less reliable, less stable, and less helpful partner.
ARTIFICIAL INTELLIGENCE
Everyone has something to say on Artificial Intelligence, and it’s all so similar that discussions about the US-China AI race are its own kind of real-life slop.
Chinese ‘embodied AI’ is primarily applying AI to robotics, while American AI focusses much more on the race to AGI (Artificial General Intelligence, human-like intellectual capacity).
Because of the way that China’s tech and media ecosystem works, many Chinese digital services and telecoms are less secure than more open markets’ counterparts. Whenever there’s a cybersecurity breach, China requires the the company to disclose it to the government. In the US, there is more public disclosure and crowdsourcing/communication between bug finders and patchers, which results in a more robust project. (In my brief search for sources to corroborate this claim, I couldn’t find anything specific, however, the logic does hold up: China does not appear to have open bug bounty platforms, whereas most major US tech giants do. You can also dive into the history of China’s short lived organized bug disclosure platform, WooYun, which was shut down in 2016. Additionally, check out this backgrounder to data ecosystems in the US, EU, and China, and this report about the Chinese hacking-for-hire ecosystem, or this one, which is probably the most comprehensive)
How the AI race plays out in each ecosystem will be interesting. People in advanced democratic economies tend to have low trust in technology and tech companies (more privacy and ethical concerns), whereas trust in technology is much higher in China. This will certainly affect the AI adoption and which use-cases are deemed appropriate in each country.
However, there still appears to be a common race to the bottom (in terms of guardrails and privacy) to try and win the AI race. Notably, the Trump Administration is not interested in regulating AI, even though American businesses’ traditional edge is the security of their products. Apple for instance, markets itself as providing most secure hardware in the industry.
One final takeaway from the AI talk was that cooperation between the US and China is generally the norm, and it still is the norm, even in today’s more competitive world. The private sectors in both countries collaborate with each other a lot, but there is not much recognition of this in the literature. Anyway, no members of any private sector company were on any of these panels, so I guess there won’t be much recognition of it in the think tank world either!
final thoughts
Now that I have all this and my bit on Critical Minerals written down, I hope I won’t have to revisit Chips/Semis, AI, or the Ally Disrespect until I learn something genuinely new.
Next post will be about something a bit more niche and hopefully funnier, but until then, I am:
Sincerely Unoriginal,
I.J. Curry