AI demand is scaling rapidly as usage expands from coding into customer service, digital platforms, and broader enterprise workflows. As workloads become more demanding, the infrastructure required to support them is also becoming significantly more complex. Hyperscale capex continues to rise as AI shifts from simple inference to agentic, long duration workloads. In our view, this points to the start of a multiyear compute supercycle — and with compute needs continuing to outpace supply, semiconductors sit at the heart of this supercycle.
Asia’s critical role in the AI supercycle
Asia sits at the core of the semiconductor ecosystem. The region holds leading positions in materials, memory, and advanced manufacturing capabilities that are difficult to replicate at scale. Japan owns critical materials and tools, Korea leads memory, and Taiwan still sets the bar in advanced fabs and packaging. Even with US onshoring, the deep supplier networks and engineering talent required for advanced nodes continue to be concentrated in Asia. These entrenched advantages create durable, high margin moats for the region’s most specialized tech leaders.