[throwup238]

Neither of which would achieve any strategic goals. TSMC is only important because Western economies crave cutting edge silicon and causing a natural disaster by destroying the dam would not only strengthen the resolve of the Chinese public but alienate the rest of the world.

[plaguuuuuu]

TSMC chips are used in basically everything military - computers obviously, but also tanks, planes, carriers, signals....

Taiwan would be quite invested in TSMC being a supplier of these.

[crote]

Are TSMC chips crucial for them, though?

As I understand it, the military generally isn't a big fan of "move fast and break things". They prefer mature, well-designed, and reliable products because it's a matter of life and death - and development takes years.

A 2025 tank isn't going to use chips made using TSMC's cutting-edge 3nm node. Personally I'd expect the fastest chips to be using at best something like the decade-old 14nm node, with the vast majority being on even more mature nodes. Sure, TSMC can produce them, but so can pretty much everyone else. Losing TSMC would be a major blow, but I doubt it'd have a huge long-term impact logistics-wise for the military.

[heylook]

> They prefer mature, well-designed, and reliable products because it's a matter of life and death - and development takes years.

First, the history of chip design and manufacturing is the history of modern weapons design[0]. As long as chips have existed we've been sticking them in brand spanking new weapons systems precisely because it's a matter of life and death. Chip makers of all kinds live and die by defense contracts.

"Because it's a matter of life and death" as a roadblock only makes sense if you don't think about it at all. If I offload some of my SIGINT processing to my brand new 3nm nodes and they don't work, won't my scheduler just mark them bad and reallocate the work back to the machines I had before? If my first shipment of 3nm drone chips doesn't deliver as promised, I've still got my existing inventory of 5nm drones and plenty of suppliers. You balance the inherent riskiness of new technology via redundancy, not avoidance.

> A 2025 tank isn't going to use chips made using TSMC's cutting-edge 3nm node.

No shit. But 2025 servers can, so can 2025 drones, so can 2025 bombs, so can 2025 satellites. Do you think whoever figures out how to communicate via quantum entanglement[1] is going to just not do it because "it's a matter of life and death"? Hell no. They're going to use it to run circles around their enemies while it's working, and they'll fall back to older communication channels when it isn't working.

[0] https://www.airandspaceforces.com/article/0310bombs/ [1] https://www.scientificamerican.com/article/china-reaches-new...

[rl3]

Deterrence is in and of itself a strategic goal.