In 2021, Taiwan experienced its worst drought in a century, leading to water restrictions and a substantial agricultural loss. The high-tech industry, also heavily dependent on water, barely managed to sustain itself with substantial government support, often at the expense of the general populace. The drought coincided with the sudden spike in chip demand during Covid, which then led to the global chip shortage. The supply crisis brought Taiwan's semiconductor industry under international scrutiny. As the pandemic accelerated our reliance on digital infrastructure and deepened the grip of cognitive capitalism.
The Semiconductor industry heavily relies on natural resources. Global landscapes from Nevada's lithium mines to Bolivia's salt flats and Inner Mongolia's mines, are constantly changing to support the accelerated development of the Semiconductor Industry1. ChatGPT, for instance, trained by NVIDIA A100, based on TSMC's 7nm technology, averages 200 million visits daily and consumes about 500,000 kWh of electricity daily2, and this is still excluding the significant energy required for model training. The ecological footprint of computation is staggering, and still expanding.
This is the core of the Taiwan dilemma: in order to survive economically and defend its sovereignty, Taiwan must continue to maintain in semiconductor manufacturing. But in doing so, it accelerates the very systems, including AI infrastructure, surveillance capitalism, ecological extraction, that are generating new crises. During the pandemic, Taiwan's then-President Tsai Ing-wen reiterated the importance of the "Silicon Shield,"3 a term used to describe how the country's chip dominance serves as both economic leverage and a deterrent against Chinese aggression. Taiwan's semiconductor capacity is not just a technical asset, it is a geopolitical instrument of sovereignty.
AI models like GPT-4 and Stable Diffusion run on NVIDIA chips fabricated by TSMC. Critics of AI rarely interrogate the geopolitical, material, and historical conditions that make AI systems possible. Taiwan is structurally entangled in its acceleration. Building the digital infrastructure also enables extractive computation, algorithmic governance, and the consolidation of power through information. The irony then comes that to protect its sovereignty, Taiwan must help accelerate a system that undermines sovereignty elsewhere. What protects the island may destabilize others. The silicon is not neutral. It underpins the tools of generative AI and militarized computing. The exact fighter jet, equipped with TSMC-manufactured chips, was deployed by the Israeli government in Gaza4.
With the passage of the U.S. and European Chip Acts and increasing restrictions aimed at China's access to high-end processors5, Taiwan has become the pivot point in an emerging techno-political order. The very infrastructure of artificial intelligence is built on decades of political engineering. Understanding TSMC's origins makes clear that advanced computation was never the result of free markets alone. It was infrastructural, strategic, and deeply historical, held together in a time of uncertainty. Today, that uncertainty hasn't disappeared.
The chips Taiwan produces no longer simply power consumer electronics; they now operate in systems of generative AI, predictive policing, algorithmic governance, and synthetic culture. Taiwan enables the global acceleration of artificial intelligence, even as it remains one of the regions most at risk if that acceleration veers toward instability or conflict. And in that demand, there is no guarantee that Taiwan itself won't be made expendable.