The AI industry chain is experiencing a divergence in market performance, with the upstream raw material price-increase segment attracting significant capital attention in the second quarter. The underlying driver is not merely short-term supply-demand imbalances for core materials, but a more fundamental structural conflict between the "slow supply" characteristic of hard-tech manufacturing and the "rapid iteration" of computing power demand, fostering a market environment of simultaneous volume and price growth.

From a supply perspective, this "slow supply" dynamic is underpinned by multiple barriers. Taking PCB substrates, high-end copper foil, and electronic glass fabric as examples, the typical lead time for core raw material capacity expansion ranges from 18 to 24 months, with most processes involving complex chemical formulations and precision rolling technologies, resulting in a slow production ramp-up. A more critical structural barrier is that outdated capacity from traditional industrial sectors cannot be easily repurposed to produce AI-grade materials. For instance, the high-frequency, high-speed copper-clad laminates required for AI servers demand extremely low dielectric loss and high heat resistance, with only a handful of manufacturers globally capable of stable mass production. This landscape—characterized by overseas monopolies, lengthy certification cycles, and scarce effective capacity—grants these companies significant pricing power and sustained profitability.

On the demand side, the explosive growth in AI computing power demand represents not just an increase in "volume" but a qualitative leap in "quality," fundamentally elevating the value proposition of upstream materials. Using MLCCs and copper-clad laminates as examples, traditional server motherboards only require standard thin electronic fabric, whereas AI servers and 400G/800G switches necessitate ultra-thin, low-loss fabric and high-frequency resins, often commanding a multiple increase in unit value. Similarly, driven by the HBM (High Bandwidth Memory) packaging boom fueled by storage demand, the consumption of advanced packaging substrates and related specialty tungsten and copper foil far exceeds that of traditional DRAM. This "technological inflation"—stemming from the shift in AI architecture from general-purpose computing to heterogeneous computing—directly triggers structural price increases upstream.

Against this backdrop of AI industry chain differentiation and the simultaneous volume-price surge in upstream core materials, Guangfa Qian Yuan Value Fund manager Wang Liyuan shifted focus towards AI computing power and new materials in Q2. The investment thesis is based on global cloud providers significantly raising AI capital expenditures, driving explosive demand for computing servers, HBM, and advanced packaging. Consequently, industry chain profits are increasingly shifting upstream to scarce raw materials. Multiple segments, including copper foil, PCB substrates, passive components, semiconductor consumables, and the tungsten supply chain, are experiencing concurrent volume and price growth, with scarce supply underpinning strong industry pricing power.

"Overall, upstream materials generally feature high technical barriers, overseas monopolies, and long capacity expansion cycles, making supply difficult to match with explosive computing power demand. This inflationary cycle is expected to persist over the medium to long term, offering significantly superior profit elasticity and earnings certainty compared to downstream segments," Wang Liyuan analyzed. The long-term growth logic for the upstream computing power sector is robust, with the tight supply-demand balance unlikely to reverse. Coupled with the ongoing release of domestic substitution benefits, the sector possesses ample growth momentum and is expected to consistently deliver excess investment returns over the medium to long term, making it a core, high-value allocation area currently. Based on this, Guangfa Qian Yuan Value Fund seized the investment opportunity in the core inflationary segments of the AI computing power upstream during the second quarter.

Since July, the AI industry chain has experienced corrections and volatility. In the Q2 report, Wang Liyuan offered two suggestions for navigating market fluctuations. First is diversification. Over-concentrating capital in a single industry direction essentially amplifies uncertainty. A more rational approach is to build a diversified portfolio, hedging risks through the low correlation between different assets and investment styles. Second is maintaining perspective. The essence of investment is capturing long-term industrial trends, not precisely timing short-term fluctuations. Using discretionary funds and maintaining patience allows for greater composure amidst market ups and downs.