Key Takeaways

  • Most leading-edge artificial intelligence semiconductors originate from TSMC fabrication plants in Taiwan, creating an acute supply chain bottleneck for Western tech and defense.
  • A kinetic conflict over Taiwan would likely destroy TSMC facilities immediately, cutting off both the United States and China from advanced silicon manufacturing nodes.
  • Nuclear deterrence and mutual economic destruction fail to prevent regional aggression when conventional capabilities fall behind adversary scale.
  • Mach Industries founder Ethan Thornton argues that mass-produced, asymmetric unmanned systems can provide tactical deterrence in the Pacific comparable to Cold War nuclear parity.
  • The United States is already engaged in an undeclared Cold War with China, requiring rapid deployment of low-cost hardware to regional allies rather than reliance on legacy platforms.

The Fragility of Advanced Silicon

The global artificial intelligence race rests on a geographic bullseye. Taiwan produces the vast majority of the world's most advanced semiconductors, concentrating global supply in TSMC fabs located directly within range of Chinese coastal artillery. Tech executives often treat this supply chain as an operational risk to manage through gradual reshoring. Ethan Thornton sees it as an existential cliff.

“As we think about AI, it is an uncomfortable truth that most of our leading edge chips come out of TSMC,” Thornton explained. “The uncomfortable truth is if that war starts TSMC is going to be blown up and the US and China will lose access to advanced nodes.”

Investors pricing multi-year compute roadmaps rarely discount the immediate elimination of leading-edge capacity. In a kinetic scenario, fabs will not simply change ownership or suffer temporary export halts. They will be destroyed. That leaves both superpowers facing an abrupt halt in hardware progress, resetting advanced computing back decades in a matter of days.

Tactical Deterrence Through Asymmetric Mass

For decades, Western defense strategy relied on two assumptions: nuclear deterrence prevents great power escalation, and intertwined trade prevents regional assault. Thornton argues both assumptions fail in practice. Economic entanglement did not stop historical territorial land grabs, and nuclear weapons cannot stop localized, sub-threshold incursions.

“It's always a dangerous position to count on mutually destruction whenever your conventional weapons are deeply behind,” Thornton said. When conventional forces lack parity or cost-effective counters, adversaries test resolve through proxy theaters and perimeter pressure.

The remedy is not building more exquisite, multibillion-dollar carrier strike groups that take decades to deliver and present high-value targets to hostile missile batteries. The counter is asymmetric mass. “My dream for for unmanned systems is a world where you have the same level of deterrence we've had on the strategic side with nuclear for many decades,” Thornton noted. “You can actually achieve with tactical through drones.” Distributing autonomous, low-cost unmanned systems across Pacific allies makes an amphibious invasion mathematically unfavorable before it begins.

The Cold War Reality

Thornton is blunt about the current geopolitical state: the conflict is not a future possibility, but an active reality. “I would say we're in a cold war with China right now,” Thornton observed. “I think we're afraid to admit it. I think fronts will unfortunately continue to open up if we can't arm our allies with with asymmetric conventional technology.”

Legacy defense procurement moves on peacetime cycles, optimizing for contractor margins rather than rapid unit production. But preventing conflict requires shipping working systems to Taiwan and regional partners right now. If conventional tactical deterrence fails, the entire commercial technology stack built on TSMC silicon collapses with it.

Why It Matters

This dynamic signals a forced reallocation of capital across both defense technology and semiconductor infrastructure. Hardware founders building unmanned systems are capturing venture rounds because military buyers need cheap, mass-manufactured hardware yesterday. At the same time, private equity and venture portfolios heavily exposed to advanced compute must recognize that physical deterrence in the Taiwan Strait dictates the long-term viability of their AI investments.