Bayesian Investor Blog

Mainland China seems committed to reasserting control over Taiwan within a few years, regardless of how much force is needed. Here’s my attempt at planning a non-disastrous scenario.

I have less expertise here than in my average blog post, so I encourage readers to question my guesses.

Blockade

In April of 2028, China launches a blockade of Taiwan. It surrounds the island with their navy, threatening to seize or sink any ship leaving Taiwan that doesn’t submit to Chinese inspections.

In addition, China demands that any aircraft leaving Taiwan must land in mainland China for inspections, before continuing to its intended destination. China threatens to bomb the runways of any airport which allows an aircraft to evade this rule.

Nothing is done to these vessels beyond inspection.

Taiwan complies with these demands.

The US responds with increased economic sanctions, but no military action.

The blockade functions as a show of force. China is, for now, only causing modest inconveniences to travel. Taiwan’s international trade declines about 5%, and air travel declines 20%.

China carefully avoids using this force to acquire semiconductors that the US prohibits shipping to China, but hints that this policy isn’t intended to be permanent.

Negotiation

China offers to relax the blockade if Taiwan agrees to phase in recognition of Beijing’s sovereignty over Taiwan.

China threatens to increasingly restrict Taiwan’s international trade if Taiwan does not agree to such an arrangement.

After six months of negotiations, they agree to the following:

• The government of Taiwan will recognize the Beijing government as being the legitimate ruler of Taiwan.
• The Beijing government will, starting immediately, get a veto over any new legislation in Taiwan.
• Except for that veto, Taiwan will be able to continue governing itself as it has been doing for 10 more years. Or maybe have a few limits on its military strategy.
• After 10 years, Beijing will assert more control over Taiwan, making few promises on any limits to such control, roughly as it did with Hong Kong.

I’ve assumed that Taiwan has a relatively weak bargaining position, because I don’t see much of a plan for it to repel a committed blockade. Taiwan and the US have options to make the blockade expensive. But Beijing sees the Taiwan government as a threat to Beijing’s legitimacy, enough so that Beijing will accept large costs if needed.

I’ve looked at drone production as a partial estimate of military power.

Here are guesses that Perplexity.ai gave me (note that these might use somewhat different categories for different countries):

Country	Estimated Annual Production (2025)
China	6,000,000+
Ukraine	2,000,000–4,000,000
Russia	1,000,000–4,000,000
Turkey	Unknown (likely 10,000s–100,000s)
United States	20,000–50,000 (military-use)
Taiwan	8,000–10,000

If the US looks like it might catch up with China, that might cause China to use force earlier while it still has an advantage in drones.

Nothing about Taiwan’s plans suggests that they will be able to put up as much of a fight as Ukraine has. Taiwan or the US could have something secret that would enable them to win a military conflict with China, but if so, it’s well hidden.

Long Term Results

Sometime near the end of that 10-year period, superintelligent AIs convince most countries, including China, to adopt a form of government that is very different from that of either Beijing or Taiwan’s current government.

Soon, Taiwan’s democracy stops looking like a threat to Beijing’s legitimacy, as whatever new form of government the AIs get them to adopt will be more secure from domestic unrest. That might be because the AI advice produces policies that are demonstrably better for most people than current policies. Or it might be because AI provides governments with more power to suppress dissent.

Alternate Scenarios

This scenario is likely to be too optimistic. It assumes more wisdom from Chinese and US leaders than recent history has given us much reason to expect.

I’ll estimate a 25% chance of a negotiated agreement without casualties.

Other possibilities include:

• Active war between mainland China and Taiwan, with the US doing little more than it has done for Ukraine. Likely an expensive partial victory for China. Chinese ships might be quite vulnerable to drones.
• US military action in response to China’s blockade. It’s unlikely this goes well for the US, which has gotten complacent and fallen way behind on possibly critical drone production. But China has little recent combat experience, so they might make bad mistakes too.
• Something involving AI surprises everyone, and I’m focusing on the wrong questions here.

Concluding Thoughts

I’ve been deliberately vague about a key issue: what happens to semiconductors that are produced in Taiwan. Beijing has been planning to gain control over Taiwan much longer than it has been concerned about access to the best semiconductors. But it wouldn’t surprise me if semiconductors soon become one of Beijing’s top reasons for using force.

Any change to who gets what GPUs is likely to have some important influence on when and how we get human-level AI. I’m pretty confused as to how GPU access will change.