Electricity and heat are the main bottlenecks for AI in 2026, driving new technologies and opportunities in the Taiwanese supply chain

Counterpoint Research identifies electricity and heat management as key bottlenecks for AI development in 2026, driving demand for high-voltage direct current (HVDC), silicon photonics, and advanced packaging solutions like panel-level packaging (PLP). The need for better thermal management is increasing the use of advanced materials such as ceramic substrates, negative thermal expansion fillers, and micro channels. Taiwan's supply chain is benefiting from a "technology substitution" trend, with local companies like Eternal Materials and Nan Pao Resins entering high-end semiconductor materials due to faster response times compared to traditional Japanese suppliers. The report also highlights structural changes and growth opportunities in high-layer count PCBs (e.g., Unimicron Technology) and the testing industry, specifically in probe cards and environmental test equipment, driven by the "shift left" testing trend for expensive AI components.