Bank of China International: Computing power infrastructure drives the mainline of AI development from zero to one, while "edge cloud resonance" leads the opportunities for storage and terminal innovation.

CICC International released a research report stating that looking forward to 2026, investment in computing power infrastructure is expected to become the driving force behind the growth of AI, and investment opportunities are expected to emerge in the "from 0 to 1" stage of the industry chain. At the same time, the increase in storage prices may persist throughout the entire year of 2026, with technologies like 4F2+CBA becoming the direction of development; Consumer electronics face both cost pressures and opportunities for innovation, and innovative AI products at the edge face a trend of both "risk" and "opportunity". CICC International's main points are as follows: Computing power infrastructure drives the main trend of AI upgrading, with industry focus on the "from 0 to 1" transformation node The rapid growth in demand for large model inference side tokens is driving CSP manufacturers to actively increase capital expenditure in computing power infrastructure construction. AIInfra, while pursuing higher computing density, faster interconnection speeds, and lower interconnection losses, also has higher requirements for PCBs, electronic fabrics, copper foils, resins, and optical interconnect architectures. The bank expects Nvidia's Rubin and RubinUltra Midplane/orthogonal backplane to be configured with M9-grade PCB/CCL solutions with high-frequency, high-speed resin + HVLP copper foil + Q fabric, and optical interconnection is expected to evolve towards OCS/CPO trends. Domestic AI chips are also actively improving computing power and interconnection cluster performance, while advanced front-end processes and packaging are also actively working together to expand production. Limited growth in bit output may lead to storage price increases throughout 2026, with storage manufacturers turning to process upgrades, high-level stacking, 4F2+CBA and other technologies According to Frost & Sullivan estimates, the global storage product market size is expected to grow from $263.3 billion in 2025 to $407.1 billion in 2029, with a CAGR of approximately 11.5%. Cloud-side and edge-side storage are the two fastest-growing sub-sectors. According to TrendForce, DRAM industry capital expenditure is expected to increase from $53.7 billion in 2025 to $61.3 billion in 2026, a YoY+14% increase; Nand industry capital expenditure is expected to increase from $21.1 billion in 2025 to $22.2 billion in 2026, a YoY+5% increase. The growth in storage bit output in 2026 is limited, and the trend of rising storage prices may persist throughout the year. Meanwhile, storage manufacturers are shifting from pure capacity expansion to process upgrades, high-level stacking, and other new technologies. Chinese domestic storage manufacturers are also actively developing the 4F2+CBA technology architecture to compete with global leading manufacturers in technology competition. The architectural changes in 4F2+CBA are expected to bring incremental changes to the supply chain. Cost pressures and opportunities for innovation coexist, focusing on structural opportunities in consumer electronics TrendForce predicts that the increase in storage prices in 2025 and 2026 will drive comprehensive costs for smartphones to increase by 8-10% and 5-7% respectively. The rise in storage prices has increased cost pressures on consumer electronics BOM, and cost transmission-based price increases may affect end-product sales. Meanwhile, structural investment opportunities continue to exist. On one hand, Apple's pricing strategy is showing a trend towards populism, with entry-level MacBook, foldable phones, smart glasses, and other products expected to boost growth in Apple's supply chain. On the other hand, innovative AI hardware features such as Bean AI phones, Quark AI glasses, Apple Apple Intelligence, etc., are gradually being implemented at the edge, providing users with innovative experiences. Key risks Overheated demand for AI triggering an industry bubble. AI technological progress is slower than expected. Excess long-term supply capacity leading to price declines. Technological changes leading to the obsolescence of existing products.