Taiwan Bets on Silicon Photonics to Power the Next AI Boom

Taiwan is positioning silicon photonics as a strategic technology for the next phase of artificial intelligence and advanced semiconductor development. The Ministry of Economic Affairs said Taiwan is well placed to expand its capabilities in copackaged optics (CPO) and precision silicon photonics chip manufacturing, supported by its strong semiconductor ecosystem.

Silicon photonics is increasingly emerging as an important optical communications technology as AI systems face growing challenges related to computing capacity, data transmission and energy consumption. The technology uses optical signals instead of conventional copper-based electrical connections to transmit data between chips, potentially improving speed while reducing heat generation and power requirements.

According to the ministry, optical transmission can lower power consumption by approximately 30% to 50%, potentially enabling faster data processing and more efficient AI applications. The technology could also help address bandwidth constraints associated with conventional electronic transmission, particularly as AI workloads require increasingly large volumes of data to move rapidly between processors and other components.

Taiwan has several structural advantages in developing this technology. The country accounts for more than half of the global outsourced semiconductor assembly and testing market and possesses a highly integrated semiconductor manufacturing ecosystem. The ministry also noted that Taiwan represents more than 90% of global advanced semiconductor manufacturing capacity for 7-nanometer and smaller process technologies.

Copackaged optics is considered particularly important because it could help overcome the approximately 1.6-terabit-per-second bandwidth limitation associated with traditional electronic transmission. By integrating optical communication technologies closer to computing components, CPO could support the higher data-transfer requirements of next-generation AI infrastructure.

The Taiwanese government has therefore incorporated silicon photonics into its New 10 Major AI Infrastructure Projects. A dedicated research and development initiative was also launched earlier this year under the A+ Enterprise Innovation R&D Program to accelerate development of critical technologies.

Industry participation is another major component of Taiwan’s strategy. The Ministry of Economic Affairs and SEMI supported more than 30 leading companies in establishing the Silicon Photonics Industry Alliance in 2024. Members include Taiwan Semiconductor Manufacturing Co, ASE Technology Holding Co, MediaTek Inc and Hon Hai Precision Industry Co.

The alliance is focused on addressing technological challenges, particularly in CPO, while helping Taiwan establish greater control over the silicon photonics supply chain.

The ministry also highlighted growing international interest. Nvidia Corp invested US$4 billion in silicon photonics technology in March, reinforcing expectations that optical connectivity will become increasingly important for AI computing. The technology was also a major discussion point at Semicon Taiwan, underscoring its rising importance across the semiconductor industry.

Chemical Commodity Price Impact

The move is expected to accelerate demand for silicon photonics components, optical communication equipment, semiconductor packaging materials and advanced electronic chemicals as Taiwan expands AI-related manufacturing. For chemical commodities tracked by ChemAnalyst, the impact is likely to be positive but commodity-specific. Higher semiconductor and advanced packaging activity could increase consumption of electronic-grade solvents, specialty chemicals, photoresists, silicon-related materials, epoxy molding compounds and high-purity gases. Stronger demand may provide upward price support, particularly for high-purity and semiconductor-grade materials with limited supply. However, commodity prices may remain stable where global production capacity is sufficient, while sustained AI infrastructure investment could create firmer medium-term demand.