Recently, I noticed a very interesting phenomenon—silicon photonics concept stocks have gone from an obscure topic last year to the hottest investment theme in 2026. Many people around me are asking whether this sector is worth paying attention to, especially after some stocks have already risen above 16 yuan, making it even more important to understand the underlying logic.

Actually, to put it simply, the popularity of silicon photonics and CPO (Co-Packaged Optics) is mainly driven by the explosion in AI computing power causing transmission bottlenecks. The traditional method of transmitting signals via copper wires is no longer viable—it's too hot, too slow, and consumes too much energy. Replacing electrical signals with light for data transmission directly solves these pain points, significantly increasing speed and reducing power consumption.

From a technical perspective, silicon photonics involves shrinking the large optical components—lasers, detectors, modulators—down to the size of microprocessors and integrating them onto silicon substrates. CPO, on the other hand, places these optical modules directly next to CPUs or GPUs, packaging them on the same board. The benefit of this approach? It can save over 30% of energy consumption, which is a huge cost optimization for data centers.

I think many people haven't fully understood why these two concepts are often linked together. To be clear, implementing CPO requires integrating optical components into chip packaging, and silicon photonics technology just happens to be able to miniaturize those large optical parts to chip level. So, silicon photonics is the core technology, and CPO is its most promising application.

Taiwan's position in this industry chain is quite critical. While US tech giants hold patents and chip design, Taiwan leverages its world-leading semiconductor manufacturing and packaging capabilities to form a complete foundry ecosystem. From wafer fabrication to packaging and optical communication components, this advantage is unmatched globally.

Regarding specific stocks, TSMC (2330) is definitely in the first tier. It’s not just a foundry; it’s also defining packaging standards for CPO. Its COUPE platform is central to the entire silicon photonics development, and the CPO packaging technology scheduled for mass production in 2026 is led by TSMC. The ability to set standards creates a significant moat.

MediaTek (6451) and ASE Technology Holding (3711) are also worth close attention. MediaTek, under the Foxconn group, was among the earliest to deploy CPO packaging, with leading technology in high-speed modules like 800G and 1.6T. ASE, as a global leader in testing and packaging, has deep collaboration with TSMC, directly entering the advanced silicon photonics packaging field.

Another key player is Advanced Optoelectronic (3363), which works closely with TSMC to develop FAU (fiber array unit) connection technology. Bringing optical signals into chips hinges on this "interface," which is critical. Many analysts believe that Advanced Optoelectronic benefits the most from this, and that logic is indeed valid.

Other companies like Browave (3163), which controls passive optical components; Lianya (3081), providing laser sources essential for CPO; and Pan-Quen (6830), with its optical path positioning and inspection tech to improve yield, are all indispensable parts of the industry chain.

The logic in the US market is somewhat different. Broadcom (AVGO) and Marvell (MRVL) hold core silicon photonics designs and communication protocols. MediaTek (CRDO) recently acquired DustPhotonics, gaining direct control of photonic integrated circuit technology, with complete solutions from 800G to 1.6T and even 3.2T. This kind of transformation story is very attractive to investors seeking flexibility.

However, I must remind everyone that investing in this sector also involves several risks. First is yield issues. CPO involves packaging optical components with chips; if any part fails, the entire expensive GPU could be scrapped. When reviewing financial reports, pay close attention to gross margin trends—if revenue increases but gross profit declines, it may indicate ongoing yield struggles.

Second is the risk of technical standards battles. The market has more than just CPO; LPO (Linear Pluggable Optical Modules) is another technology. LPO is like an enhanced version of traditional pluggable modules—more power-efficient, cheaper than CPO, and easier to maintain. Before 1.6T becomes mainstream, LPO might capture a significant market share.

Third is to look at actual revenue figures. Some companies claim to be silicon photonics concept stocks but have very low optical communication-related revenue. Be cautious whether they are just riding the hype. Genuine investment support should come from major players like NVIDIA and Broadcom with real orders.

Fourth are geopolitical and policy risks. US broadband infrastructure plans will influence optical communication demand, and delays could directly impact downstream industries. Also, US restrictions on advanced semiconductor equipment could affect silicon photonics, as it’s a cutting-edge technology vulnerable to tech wars.

My view is that silicon photonics and CPO are not short-term themes but structural growth trends over the next 5 to 10 years. 2026 marks a critical turning point from R&D validation to large-scale mass production, truly testing each company's ability to implement the technology. Many silicon photonics concept stocks have already risen above 16 yuan, but future opportunities depend more on who can achieve real mass production.

In simple terms, the investment logic should be: US stocks focus on "standard-setting," while Taiwan stocks focus on "supply chain performance." While capital chases new themes, the most important thing is to return to fundamentals—prioritize companies that have been certified by major manufacturers and show a clear increase in optical communication revenue. Only then can investors avoid noise and identify truly valuable companies in this fast-moving sector.