Interesting article. Well thought out, and well researched.
First, what I liked:
- The decoupling dashboard. It's an excellent way to illustrate the state of affairs and you provided necessary and sufficient information.
- The critical window. It seems like you nailed the big hitters, and the timeline.
- Calling out that China has been preparing for the decoupling, and painting the timeline.
However, your risk analysis seems willfully overconfident.
- It is disingenuous to treat "The US and its allies" as a unified bloc while neglecting that the US has been estranging its putative allies while China has been making deals with them. This seems likely to continue.
- Belt and Road has been successful in South America, and many Latin American countries have begun to migrate from the US to China as their major trading partner for the past decade. Indeed, you point to an inability for China to feed itself without US soybeans but it has been doing just that voluntarily this year due to trade agreements with Brazil - what you claim to be a liability for China turns out to be a major headache for the US without any pain on the Chinese side. (This begs the question of how many of the other claims would result in a similar outcome?)
- The analysis neglects the US debt problem and the real risk of inflation. You mention four things that international currencies require, but then describe a strategy whereby the US violates one or more of those with IEEPA. Doing so would put dollar dominance at risk, risking real ecosystem collapse in the US. My analysis suggests doing so would be far worse for the US than for China.
- The appeals to game theory and physics are overblown. Your analysis does not contain a real treatment of either of these things, and invoking them drastically weakens your argument.
These items notwithstanding, I learned from this essay, as with all of your articles. Please treat this critique as an invitation to debate and clarify our understanding rather than an attack on your work. Thank you for writing it.
I think this article is missing a lot of important factors that break the underlying thesis. I lived in China for almost a decade and speak Mandarin, so I know a bit more about this than OP I think.
Banking equity: This is irrelevant when all banks are state owned and the government is willing and able to absorb losses in bad banks. They have done this in the past. You have to consider total government revenues as rescue capacity, not bank equity.
Taiwan: China considers this an existential issue. The US does not. If China was willing to make a $1 trillion purchase commitment for soybeans in exchange for the US withdrawing military support for Taiwan, do you think Trump (or any other isolationist) won't take this offer? I'm almost certain they would.
Chaos theory: US allies should be equal skeptical of chaotic Trump. There is no nearshoring going on because maybe next week Trump will threaten to invade Guatemala. If I'm India, do I consider Trump a reliable partner? India was sanctioned for buying Russian oil, but China was not. If the US is happy abandoning NATO, why would SK/Japan rely on US security guarantees?
The chaos theory argument only works if you're chaotic to your enemies and reliable to your friends. It's stupid for Canada to be expanding it's surveillance capabilities because they never know when the US might abandon intelligence sharing. But that's exactly what's happening, so our allies spend for a million China contingencies AND a million US contingencies.
Supply chain interdependence: This is my industry. You can't hand-wave away the fact that we no longer produce plastic injection molding in the US. Imagine if tomorrow we lose all supply of medical syringes and have no way of building molds that produce these syringes either. People die in this scenario. Plastic doesn't just mean happy meal toys, it has tons of applications in food, medical, electronics... practically every industry. Reshoring is a complicated exercise and the administration is making it much worse by alienating countries that are both friendly and better than us at producing certain things.
Demographic cliff: Your population growth charts for the US are based on historical levels of immigration. Currently we are at net negative immigration, so our population will also decline without a sudden surge in native births.
My best guess is that China and the US will both muddle along and experience low growth for completely self inflicted reasons. But they are also both large and diversified enough that there will be no Great Depression 2.0
If China was willing to make a $1 trillion purchase commitment for soybeans in exchange for the US withdrawing military support for Taiwan, do you think Trump (or any other isolationist) won't take this offer?
the answer is NO, and ABSOLUTELY NO, the ONLY way US will agree is China agrees Taiwan maintain her democratic systems untouched and PERMANENTLY, which China will never dare to agree
Agree completely regarding the chaos theory. I'm a U.S. international trade lawyer and guess what, it's not the Chinese who are paying our massive bills...
They already ditched our soybeans completely, also, a lot of the rest of the injection molding happens in Taiwan. The best stuff actually. Presumably we would lose that too. It’ll be hard to reindustrialize when 99% of our small tools and a huge portion of our fasteners come from china. Our software infrastructure systems are riddled with vulnerabilities. It’ll suck if huge parts of the country lack utilities at various times. Are we going to get anything done if pipelines are randomly shutdown?
Looks to me you are too optimistic on the short to medium term pain in US, etc. Too pessimistic on China. Where is Russia as a China friend?
What happens to China:
Month 1:
RMB collapses 30-50% (capital flight unstoppable). Comment: Really? Why? If it was unstoppable, it would have happened.
$2.5T in US assets frozen → $1T+ in losses realized. Comment: Are you sure these numbers are correct? What is the source?
Food shortages (30% of soybeans from US, can’t replace immediately). Comment: Who cares about Soy. They can eat rice, plant more soy, get from Brazil, etc. It's not a must.
Year 1:
Social unrest (property owners + unemployed workers). Comment: Naah, not unrest.
The numbers are all best attempts as making estimates for deeply uncertain/unknowable things. That's part of my process, make good faith estimates then do the backward induction. Revise them in a quasi-Bayesian way as neccessary.
I was going to write the same thing. I badly want to absorb the material because the points are interesting/provocative but the GPT-speak is so incredibly distracting.
seconding this. When reading a unique take (as opposed to listicles and tutorials), I like having that human touch. I don’t want to feel like i’m asking chatgpt for an analysis.
interesting, feel like i have to agree with many points, but feel it is too one sided trying to point china as being weak. You mention the usa can achieve self sufficiency on rare earth in 10y time... how certain are you about this? what do you base this on?
Great essay. Although you should know the UN population projections you rely on assume China TFR will rebound to 1.5 and stay in the mid 1s for much of the century.
This seems very unlikely to me. If they stay at current levels, their population is sub 400m by 2100. And the dependency ratios are much worse and get worse faster.
I got linked to your post from Marginal Revolution. Why not just share the prompts you used? You should at least put a "slop" warning. No one wants to read slop they didn't generate themselves.
Economic power follow military might. Going with the current trend China navy will be more powerful than the US navy in 5 years, infinity much powerful in 10 years.
And there is literally nothing the U.S. can do about it.
You can seize 2 trillion from China, but it does not matter, if they seize Hawái in answer
A combination of AI and robotics will allow China to continue to grow its economy and by extension, geopolitical power and influence, well into the future, despite demographic headwinds.
While it's true that at the leading edge of AI development, Chinese AI models are slightly behind American ones (e.g. Alibaba's Qwen 3 Max LLM is ranked #3 on LMArena's Text leaderboard), the country has key advantages when it comes to AI and robotics.
* "I'm not saying China technologically is winning in the model war, but in terms of the actual application and also people benefiting from AI, it has made a lot of development.
* I've seen a survey of Chinese firms. Last year, only 8% were using AI in their business. Now, that number is approaching half. So the adoption rate has been fairly fast."
* "The [Chinese] government is all in, embracing it [AI]. In fact, the government, just a few weeks ago, launched a kind of AI+ policy. They say in 2030, which is 5 years from now, they want to see 90% penetration of AI agents and devices and things like that in society."
This all-out embrace of AI is coming at a time when the youth unemployment rate is 18%. In other words, to Beijing, AI adoption and proliferation is more important than providing job opportunities for young people. Why? Beijing knows it must overcome its demographics issues.
But it is in robotics where China's advantages really shine. At the end of the day, success in robotics comes down to just two huge factors.
1. Manufacturing capacity
China accounts for >30% of global manufacturing. Despite tariffs, "China’s overall trade surplus is on track to exceed $1trn this year".
The same politicians who put up trade barriers want Chinese investment: "From Brasília to Berlin and Bangkok, politicians hear calls to protect established industries from Chinese competition. Yet many of the same politicians want Chinese investors to help them build industries of the future, by opening plants to make batteries, say. That limits their desire to confront China."
Meanwhile, "China adds more electricity demand than the entire annual consumption of Germany, every single year."
Given that the critical inputs to AI are chips and electricity, even when forced to use inferior chips, China can continue to advance its AI by using more electricity to overcompensate for worse chips.
Alex says, "You can’t run a manufacturing export economy with a shrinking workforce." China's dark factories, where robots don't even need lights to work, would beg to differ. https://www.youtube.com/shorts/Z2HwRrrNQ1w
If a startup team of 5 using AI can have the same level of productivity as a team of 50 without AI, then why can't 880m working-age Chinese in 2035, empowered by AI, robotics, and cheap and abundant energy, produce the output of a workforce several times larger?
China's property slump
Alex has pointed out that China has a massive oversupply of housing and is going through a terrible property slump. It's clear that this really hurts asset owners, i.e. the old and rich. But who benefits from low housing costs? The young and poor, who do not have assets.
If young people can live where they want without paying an arm and a leg for rent/mortgage, does that help or hurt the overall rate of development and innovation?
When housing prices increase, that is a transfer of wealth from young, poor people to old, rich people. When housing prices fall or plateau while the economy grows, it is the reverse.
Trump’s style may prevail in the first game. The problem is that there will be multiple rounds. One factor not accounted for here is the asymmetry in each society’s capacity to “eat bitter”. Not so sure that the US has the upper hand there. Finally this analysis totally ignores our dependence on China for the little subcomponents that we need to make, well, just about everything.
Excellent article. My only question is why are you certain that decoupling will continue long-term?
"I’m observing that certain futures are impossible (re-integration), certain futures are inevitable (continued separation)"
What if, for instance, a new Democrat president in 2029 is hugely pro-free-trade, globalisation, perhaps even pro-China for personal/professional reasons. Is it that much of a stretch to imagine a détente between the U.S. and China, resulting in partial re-integration?
I'm not suggesting this scenario is likely, but surely it's plausible?
I actually think that while Trump's cluelessness and increasing dementia has this crazy ivan impact on the Chinese, he has proven that the Chinese can out negotiate him. I expect that for money, he is letting Nividia increasing its chips sales over time and Trump won't really do much if Taiwan is attacked. If a Democrat comes into office, they will be more stable but harder on China. Trump is a pushover for the Chinese.
Interesting article. Well thought out, and well researched.
First, what I liked:
- The decoupling dashboard. It's an excellent way to illustrate the state of affairs and you provided necessary and sufficient information.
- The critical window. It seems like you nailed the big hitters, and the timeline.
- Calling out that China has been preparing for the decoupling, and painting the timeline.
However, your risk analysis seems willfully overconfident.
- It is disingenuous to treat "The US and its allies" as a unified bloc while neglecting that the US has been estranging its putative allies while China has been making deals with them. This seems likely to continue.
- Belt and Road has been successful in South America, and many Latin American countries have begun to migrate from the US to China as their major trading partner for the past decade. Indeed, you point to an inability for China to feed itself without US soybeans but it has been doing just that voluntarily this year due to trade agreements with Brazil - what you claim to be a liability for China turns out to be a major headache for the US without any pain on the Chinese side. (This begs the question of how many of the other claims would result in a similar outcome?)
- The analysis neglects the US debt problem and the real risk of inflation. You mention four things that international currencies require, but then describe a strategy whereby the US violates one or more of those with IEEPA. Doing so would put dollar dominance at risk, risking real ecosystem collapse in the US. My analysis suggests doing so would be far worse for the US than for China.
- The appeals to game theory and physics are overblown. Your analysis does not contain a real treatment of either of these things, and invoking them drastically weakens your argument.
These items notwithstanding, I learned from this essay, as with all of your articles. Please treat this critique as an invitation to debate and clarify our understanding rather than an attack on your work. Thank you for writing it.
totally agreed
I think this article is missing a lot of important factors that break the underlying thesis. I lived in China for almost a decade and speak Mandarin, so I know a bit more about this than OP I think.
Banking equity: This is irrelevant when all banks are state owned and the government is willing and able to absorb losses in bad banks. They have done this in the past. You have to consider total government revenues as rescue capacity, not bank equity.
Taiwan: China considers this an existential issue. The US does not. If China was willing to make a $1 trillion purchase commitment for soybeans in exchange for the US withdrawing military support for Taiwan, do you think Trump (or any other isolationist) won't take this offer? I'm almost certain they would.
Chaos theory: US allies should be equal skeptical of chaotic Trump. There is no nearshoring going on because maybe next week Trump will threaten to invade Guatemala. If I'm India, do I consider Trump a reliable partner? India was sanctioned for buying Russian oil, but China was not. If the US is happy abandoning NATO, why would SK/Japan rely on US security guarantees?
The chaos theory argument only works if you're chaotic to your enemies and reliable to your friends. It's stupid for Canada to be expanding it's surveillance capabilities because they never know when the US might abandon intelligence sharing. But that's exactly what's happening, so our allies spend for a million China contingencies AND a million US contingencies.
Supply chain interdependence: This is my industry. You can't hand-wave away the fact that we no longer produce plastic injection molding in the US. Imagine if tomorrow we lose all supply of medical syringes and have no way of building molds that produce these syringes either. People die in this scenario. Plastic doesn't just mean happy meal toys, it has tons of applications in food, medical, electronics... practically every industry. Reshoring is a complicated exercise and the administration is making it much worse by alienating countries that are both friendly and better than us at producing certain things.
Demographic cliff: Your population growth charts for the US are based on historical levels of immigration. Currently we are at net negative immigration, so our population will also decline without a sudden surge in native births.
My best guess is that China and the US will both muddle along and experience low growth for completely self inflicted reasons. But they are also both large and diversified enough that there will be no Great Depression 2.0
If China was willing to make a $1 trillion purchase commitment for soybeans in exchange for the US withdrawing military support for Taiwan, do you think Trump (or any other isolationist) won't take this offer?
the answer is NO, and ABSOLUTELY NO, the ONLY way US will agree is China agrees Taiwan maintain her democratic systems untouched and PERMANENTLY, which China will never dare to agree
Agree completely regarding the chaos theory. I'm a U.S. international trade lawyer and guess what, it's not the Chinese who are paying our massive bills...
They already ditched our soybeans completely, also, a lot of the rest of the injection molding happens in Taiwan. The best stuff actually. Presumably we would lose that too. It’ll be hard to reindustrialize when 99% of our small tools and a huge portion of our fasteners come from china. Our software infrastructure systems are riddled with vulnerabilities. It’ll suck if huge parts of the country lack utilities at various times. Are we going to get anything done if pipelines are randomly shutdown?
Intriguing. Thank you.
Looks to me you are too optimistic on the short to medium term pain in US, etc. Too pessimistic on China. Where is Russia as a China friend?
What happens to China:
Month 1:
RMB collapses 30-50% (capital flight unstoppable). Comment: Really? Why? If it was unstoppable, it would have happened.
$2.5T in US assets frozen → $1T+ in losses realized. Comment: Are you sure these numbers are correct? What is the source?
Food shortages (30% of soybeans from US, can’t replace immediately). Comment: Who cares about Soy. They can eat rice, plant more soy, get from Brazil, etc. It's not a must.
Year 1:
Social unrest (property owners + unemployed workers). Comment: Naah, not unrest.
Regional governments defy Beijing (withhold revenue). Comment: Impossible.
What happens to US:
Month 1 to Year 1: Comment: I largely agree but the pain will be greater!
____
Agree long-term though.
The numbers are all best attempts as making estimates for deeply uncertain/unknowable things. That's part of my process, make good faith estimates then do the backward induction. Revise them in a quasi-Bayesian way as neccessary.
Cool. Thank you Alex ;-)
Count the number of times this article uses the same rhetorical device: "This isn't just X. It's Y."
I'm fine with AI writing but can you do mix it up a little, please?
For example:
-decoupling isn’t just inevitable, but asymmetric
-It’s not a bug. It’s the feature that makes everything else inevitable.
-That’s not a solvency problem—it’s a physics problem.
-Trump’s “madman” strategy isn’t madness—it’s mathematically optimal.
-Because the trade of our generation isn’t predicting which future arrives. It’s observing that certain futures are impossible
-We’re not halfway through this process. We’re barely a quarter done.
-Not in some abstract geopolitical sense. In cold, hard, mathematical reality.
-Not in some abstract geopolitical sense. In cold, hard, mathematical reality.
-The property bubble didn’t just pop. It revealed that the emperor has no clothes.
-The question isn’t whether decoupling happens. The question is whether China survives it.
-The property crisis isn’t a bug in their economic model. It was the model.
-This isn’t a solvency problem. It’s a physics problem.
-This isn’t just about GDP growth or bank balance sheets. It’s about regime survival.
-The CCP isn’t worried about a recession. They’re worried about regime collapse.
-This wasn’t random protectionism. This was prep for decoupling.
This would have been soooo much better if it wasn’t obviously so deeply edited by ChatGPT
I was going to write the same thing. I badly want to absorb the material because the points are interesting/provocative but the GPT-speak is so incredibly distracting.
seconding this. When reading a unique take (as opposed to listicles and tutorials), I like having that human touch. I don’t want to feel like i’m asking chatgpt for an analysis.
1990. China's economy has come to a halt. The Economist
1996. China's economy will face a hard landing. The Economist
1998. China's economy’s dangerous period of sluggish growth. The Economist
1999. Likelihood of a hard landing for the Chinese economy. Bank of Canada
2000. China currency move nails hard landing risk coffin. Chicago Tribune
2001. A hard landing in China. Wilbanks, Smith & Thomas
2002. China Seeks a Soft Economic Landing. Westchester University
2003. Banking crisis imperils China. New York Times
2004. The great fall of China? The Economist
2005. The Risk of a Hard Landing in China. Nouriel Roubini
2006. Can China Achieve a Soft Landing? International Economy
2007. Can China avoid a hard landing? TIME
2008. Hard Landing In China? Forbes
2009. China's hard landing. China must find a way to recover. Fortune
2010: Hard landing coming in China. Nouriel Roubini
2011: Chinese Hard Landing Closer Than You Think. Business Insider
2012: Economic News from China: Hard Landing. American Interest
2013: A Hard Landing In China. Zero Hedge
2014. A hard landing in China. CNBC
2015. Congratulations, You Got Yourself A Chinese Hard Landing. Forbes
2016. Hard landing looms for China. The Economist
2017. Is China's Economy Going To Crash? National Interest
2018. China's Coming Financial Meltdown. The Daily Reckoning.
2019 China's Economic Slowdown: How worried should we be? BBC2020. Coronavirus Could End China's Decades-Long Economic Growth Streak. NY Times
2021 Chinese economy risks deeper slowdown than markets realize. Bloomberg
2022. China Surprise Data Could Spell R-e-c-e-s-s-i-o-n. Bloomberg.
2023. No word should be off-limits to describe China's faltering economy. Bloomberg
2024. China Slowdown Means It May Never Overtake US ...Bloomberg
Looks like this bullshit never goes out of style..
interesting, feel like i have to agree with many points, but feel it is too one sided trying to point china as being weak. You mention the usa can achieve self sufficiency on rare earth in 10y time... how certain are you about this? what do you base this on?
rare earths aren't that rare
yes but the whole infrastructure to extract them? often by products of other industries.
Can take many years of capex to get a mine started + operating a full capacity, and you need many many mines.
No country on earth has the refining capacity that china does (+90% global production)
US may have a few more critical dependencies? like API's and pharma precursors
Pharma is a good point, I should incorporate that into subsequent versions, thanks for highlighting.
Great essay. Although you should know the UN population projections you rely on assume China TFR will rebound to 1.5 and stay in the mid 1s for much of the century.
This seems very unlikely to me. If they stay at current levels, their population is sub 400m by 2100. And the dependency ratios are much worse and get worse faster.
https://distantprovince.by/posts/its-rude-to-show-ai-output-to-people/
I got linked to your post from Marginal Revolution. Why not just share the prompts you used? You should at least put a "slop" warning. No one wants to read slop they didn't generate themselves.
delusional china is doomed slop recycled for past 15 yrs.
The amount of copium is truly astounding.
LOL so much wishful thinking
Economic power follow military might. Going with the current trend China navy will be more powerful than the US navy in 5 years, infinity much powerful in 10 years.
And there is literally nothing the U.S. can do about it.
You can seize 2 trillion from China, but it does not matter, if they seize Hawái in answer
Why China CAN win
A combination of AI and robotics will allow China to continue to grow its economy and by extension, geopolitical power and influence, well into the future, despite demographic headwinds.
While it's true that at the leading edge of AI development, Chinese AI models are slightly behind American ones (e.g. Alibaba's Qwen 3 Max LLM is ranked #3 on LMArena's Text leaderboard), the country has key advantages when it comes to AI and robotics.
Alibaba chairman Joe Tsai at All-In Summit 2025: https://www.youtube.com/watch?v=sp35RmulTDc&t=1127s
* "I'm not saying China technologically is winning in the model war, but in terms of the actual application and also people benefiting from AI, it has made a lot of development.
* I've seen a survey of Chinese firms. Last year, only 8% were using AI in their business. Now, that number is approaching half. So the adoption rate has been fairly fast."
* "The [Chinese] government is all in, embracing it [AI]. In fact, the government, just a few weeks ago, launched a kind of AI+ policy. They say in 2030, which is 5 years from now, they want to see 90% penetration of AI agents and devices and things like that in society."
This all-out embrace of AI is coming at a time when the youth unemployment rate is 18%. In other words, to Beijing, AI adoption and proliferation is more important than providing job opportunities for young people. Why? Beijing knows it must overcome its demographics issues.
But it is in robotics where China's advantages really shine. At the end of the day, success in robotics comes down to just two huge factors.
1. Manufacturing capacity
China accounts for >30% of global manufacturing. Despite tariffs, "China’s overall trade surplus is on track to exceed $1trn this year".
The same politicians who put up trade barriers want Chinese investment: "From Brasília to Berlin and Bangkok, politicians hear calls to protect established industries from Chinese competition. Yet many of the same politicians want Chinese investors to help them build industries of the future, by opening plants to make batteries, say. That limits their desire to confront China."
Source: https://www.economist.com/international/2025/09/23/a-made-in-china-plan-for-world-domination
2. Electricity supply
Despite comparable levels of total economic output, China generates 2.3x more electricity than the US: https://www.visualcapitalist.com/ranked-top-countries-by-annual-electricity-production-1985-2024/
The lack of electricity in the US is a major bottleneck in developing AI: https://fortune.com/2025/08/14/data-centers-china-grid-us-infrastructure/
Meanwhile, "China adds more electricity demand than the entire annual consumption of Germany, every single year."
Given that the critical inputs to AI are chips and electricity, even when forced to use inferior chips, China can continue to advance its AI by using more electricity to overcompensate for worse chips.
Alex says, "You can’t run a manufacturing export economy with a shrinking workforce." China's dark factories, where robots don't even need lights to work, would beg to differ. https://www.youtube.com/shorts/Z2HwRrrNQ1w
If a startup team of 5 using AI can have the same level of productivity as a team of 50 without AI, then why can't 880m working-age Chinese in 2035, empowered by AI, robotics, and cheap and abundant energy, produce the output of a workforce several times larger?
China's property slump
Alex has pointed out that China has a massive oversupply of housing and is going through a terrible property slump. It's clear that this really hurts asset owners, i.e. the old and rich. But who benefits from low housing costs? The young and poor, who do not have assets.
If young people can live where they want without paying an arm and a leg for rent/mortgage, does that help or hurt the overall rate of development and innovation?
When housing prices increase, that is a transfer of wealth from young, poor people to old, rich people. When housing prices fall or plateau while the economy grows, it is the reverse.
Trump’s style may prevail in the first game. The problem is that there will be multiple rounds. One factor not accounted for here is the asymmetry in each society’s capacity to “eat bitter”. Not so sure that the US has the upper hand there. Finally this analysis totally ignores our dependence on China for the little subcomponents that we need to make, well, just about everything.
Excellent article. My only question is why are you certain that decoupling will continue long-term?
"I’m observing that certain futures are impossible (re-integration), certain futures are inevitable (continued separation)"
What if, for instance, a new Democrat president in 2029 is hugely pro-free-trade, globalisation, perhaps even pro-China for personal/professional reasons. Is it that much of a stretch to imagine a détente between the U.S. and China, resulting in partial re-integration?
I'm not suggesting this scenario is likely, but surely it's plausible?
I actually think that while Trump's cluelessness and increasing dementia has this crazy ivan impact on the Chinese, he has proven that the Chinese can out negotiate him. I expect that for money, he is letting Nividia increasing its chips sales over time and Trump won't really do much if Taiwan is attacked. If a Democrat comes into office, they will be more stable but harder on China. Trump is a pushover for the Chinese.