Open Source Securities: AI is the main focus while considering investment opportunities in multiple sectors. It is recommended to pay attention to sectors such as telecommunications operators and AI applications.

Open Source Securities released a research report predicting that in 2026, the "siphon effect" of AI will be significant, and global AI may continue to resonate. On the international front, giants like Google are continuously increasing their investment in AI capital, guiding investments in Google Gemini and other large model Tokens, and the positive feedback loop of AI is gradually becoming more prominent. Domestically, Chinese AI giants may enter a period of large-scale investment in AI computing power. The firm is bullish on the five core themes of "light, fiber optic, liquid cooling, domestic AI, satellite," and also recommends paying attention to sectors such as telecommunications operators and AI applications. Key points from Open Source Securities include: 2026 Track One: Global AI resonance, focusing on the four main themes of "light, fiber optic, liquid cooling, domestic AI" In 2026, it may become the year of mass production for 1.6T optical modules, with silicon photonics technology accelerating its deployment and potentially becoming the mainstream solution for high-speed optical modules. This may drive a significant increase in demand for optical chips such as EML and CW lasers, as well as the "AIDC + drones + DCI" triad resonance continuing to drive demand for optical fiber cables. New technologies such as OCS/CPO/ Hollow-core optical fiber/thin film lithium niobate/LPO/LRO are also expected to present new investment opportunities. In 2026, with the gradual shipping of NVIDIA's GB300 NVL72, Rubin, and Rubin CPX series server cabinets, the increasing shipments of new computing chips by giants such as Google, as well as the scale shipping of domestic computing chips and the gradual increase in production of domestic super nodes, the penetration rate of liquid cooling may significantly increase. With the joint promotion of national policies and domestic giants, the domestic AI industry chain is expected to thrive in 2026, and it is recommended to focus on computing, networking, AIDC, and other core directions. 2026 Track Two: Domestic AI and satellites collaborate to build a space, air, and ground computing network The demand for large model training is continuously growing, and space has unique advantages such as infinite CECEP Solar Energy and natural vacuum cooling environment with global seamless coverage, making it the core direction for the development of the next generation of computing power. SpaceX plans to deploy a huge space computing constellation of millions of satellites, and companies like Blue Origin, Google, and Starcloud are also making moves in this area. China attaches great importance to the construction of a space, air, and ground computing network, where satellites serve as the core cornerstone of the space network, and it is recommended to focus on four key areas: (1) satellite manufacturing: benefiting from the increasing demand for satellite launches, focusing on satellite payload suppliers and satellite platform component suppliers; (2) satellite launch: focusing on the industrial catalysis brought about by the allocation of launch resources, launch pace, and technological development; (3) ground equipment: focusing on the construction of the high-value core network, and the terminal market; (4) satellite operations: Starlink and Yuanxin respectively lead the star network, G60 jointly advancing, expected to quickly build the satellite network. 2026 Track Three: AI applications surging, opportunities for operators, modules, controllers, CDNs, and more In the AI era, operators are vigorously developing AI, and AI cloud is expected to become a new engine for performance growth. The advancement of technologies such as voice recognition and image recognition based on AI models has accelerated the development of AI toys. AI toys, through the combination of intelligent computing modules, cellular/Wi-Fi modules, and other hardware, have achieved human-like interaction capabilities. With the continuous optimization of edge AI models, the demand for intelligent computing modules and controllers may gradually increase. AI models are the basis for embodied intelligence development, and pruned and quantified small models perform well at the edge terminals. The demand for intelligent computing modules and controllers may gradually increase. AI applications are expected to generate demand for millisecond-level response times, and CDNs are expected to push computing power to user terminals through the layout of edge node networks, achieving localization and lightweight processing of data, thereby continuously driving demand for CDNs. Risk warning: AI development falls short of expectations, satellite development falls short of expectations, escalating trade tensions between China and the US, etc.