> If China can already make good chips (or will very soon), then export controls haven’t slowed them down in any meaningful way. In that world, we should ship Nvidia chips over as fast as we can fab them.
> If these hypothetical new fabs come online (say, ~12 months), China can invade or blockade Taiwan without worrying about losing TSMC.
I know there are a lot of reasons why China wants Taiwan, but if China develops its own domestic chip production, could that also be somewhat of a disincentive to invade/blockade Taiwan? As in, they could reduce invasion incentives by removing the strategic incentive to control semiconductor manufacturing in order to win the race to ASI?
If China believes winning the AI race is the biggest priority given how it could fundamentally reshape the world order, then I think p(Taiwan invasion) decreases because (1) it strains resources that should be going towards supporting domestic AI development + distracts from the overarching goal, (2) it doesn't offer strategic advantage if China is able to make its own good chips, and (3) if China achieves ASI first it could take over the whole world, not just Taiwan.
(1) - Ah the idea is that if it is coming in the next year, building dependence on Chinese chips is the way that we can compete in their domestic market (eg. reducing SMIC revenues). I do think that, to the extent that this is an inevitability / a state target that is indifferent to profit, this is much harder
(2) - I guess implicitly assumed in my argument is that the TSMC fabs get destroyed + the TSMC workers all leave if a Taiwan invasion (either by Taiwan, which may have credibly committed to do so in the event of an invasion to act as a disincentive, or maybe by the US).
Oh that's a very interesting point re. limited resources (eg. you might think that the vast global backlash on China / global recession may be a huge downer on domestic AI capabilities). I think this is probably counteracted by the fact that, if China is taking AGI seriously, it can slow the US down (who, under this view, have a much weaker domestic supply chain) and that might be worth it to them.
Great post! Is it possible we should pursue (4) onshoring the rest of the supply chain) anyway? Why couldn’t we double down on lobbying our allies in Europe to not sell parts to China?
Yes definitely -- let the great reindustrialization project begin. Getting the secondary export controls may be hard given the power of Nvidia's lobbying, but I do agree, we should, independent of whether China can make good chips or not...
> If China can already make good chips (or will very soon), then export controls haven’t slowed them down in any meaningful way. In that world, we should ship Nvidia chips over as fast as we can fab them.
I don't fully understand the above argument on its own (like without the "even better" part after it, which implies it should stand alone). If compute is scarce globally (including for US companies), why would we want to increase competition for limited compute by intentionally selling to China? Especially memory chips-- https://www.bloomberg.com/news/articles/2025-11-26/tech-firms-from-dell-to-hp-warn-of-memory-chip-squeeze-from-ai
> If these hypothetical new fabs come online (say, ~12 months), China can invade or blockade Taiwan without worrying about losing TSMC.
I know there are a lot of reasons why China wants Taiwan, but if China develops its own domestic chip production, could that also be somewhat of a disincentive to invade/blockade Taiwan? As in, they could reduce invasion incentives by removing the strategic incentive to control semiconductor manufacturing in order to win the race to ASI?
If China believes winning the AI race is the biggest priority given how it could fundamentally reshape the world order, then I think p(Taiwan invasion) decreases because (1) it strains resources that should be going towards supporting domestic AI development + distracts from the overarching goal, (2) it doesn't offer strategic advantage if China is able to make its own good chips, and (3) if China achieves ASI first it could take over the whole world, not just Taiwan.
Would love to know your thoughts!! :)
(1) - Ah the idea is that if it is coming in the next year, building dependence on Chinese chips is the way that we can compete in their domestic market (eg. reducing SMIC revenues). I do think that, to the extent that this is an inevitability / a state target that is indifferent to profit, this is much harder
(2) - I guess implicitly assumed in my argument is that the TSMC fabs get destroyed + the TSMC workers all leave if a Taiwan invasion (either by Taiwan, which may have credibly committed to do so in the event of an invasion to act as a disincentive, or maybe by the US).
Oh that's a very interesting point re. limited resources (eg. you might think that the vast global backlash on China / global recession may be a huge downer on domestic AI capabilities). I think this is probably counteracted by the fact that, if China is taking AGI seriously, it can slow the US down (who, under this view, have a much weaker domestic supply chain) and that might be worth it to them.
Agree with the ASI point, see above.
Great post! Is it possible we should pursue (4) onshoring the rest of the supply chain) anyway? Why couldn’t we double down on lobbying our allies in Europe to not sell parts to China?
Yes definitely -- let the great reindustrialization project begin. Getting the secondary export controls may be hard given the power of Nvidia's lobbying, but I do agree, we should, independent of whether China can make good chips or not...