Chinese GPU/chip advancement: multiple pathways to 'strength through use' in domestic chip adoption.

Multiple positive developments have boosted AI investment confidence, with China's domestic AI industry chain-focused Huabao Shanghai Stock Exchange Innovation Board AI ETF (ticker: 589520) rising as much as 9.56% on July 31st, the final trading day of the month, closing up 6.73%. Data shows the ETF attracted net inflows for five consecutive days totaling 33.94 million yuan, and over a 20-day trading period accumulated inflows of 80.63 million yuan.

Among the fund's holdings, Orbbec (Orbbec-W) led the gains with an increase exceeding 17%, UCloud (UCloud-W) rose over 15%, while Chipown Semiconductor and Helixtech Information gained over 13%. Among heavyweight positions, Cambrian and Kingsoft Office rose over 6%, Lansheng Technology rose over 4%, and Mosixc (Mosixc-U) rose over 2%.

The strong rebound of China's domestic AI industry chain stems from both international and domestic factors. Internationally, SK Hynix Chairman Choi Tae-won purchased additional SK Hynix common shares, Amazon announced increased capital expenditure for the year, and Microsoft delivered better-than-expected earnings, collectively restoring AI investment confidence. Domestically, top-level policy meetings confirmed implementation of the "AI+" initiative, while China's National Development and Reform Commission indicated that computing power network construction will drive 4 trillion yuan in new investment.

As a core frontier technology, the importance of self-sufficiency and control over AI has become increasingly evident. Renowned economist Ren Zeping identified three major bottlenecks in China's domestic AI sector: equipment, manufacturing, and software ecosystems. Advanced process nodes are constrained by EUV lithography limitations, with domestic manufacturing capped at roughly 7nm equivalence—a two-generation gap. Additionally, NVIDIA's CUDA ecosystem has accumulated 20 years of development, while China's software stack lags by 5–10 years, making ecosystem catch-up a key priority.

Ren Zeping outlined three breakthrough pathways for domestic GPU development, enabling China's chips to grow stronger through greater adoption: ① Architectural innovation, exemplified by DeepSeek's Mixture of Experts (MoE) architecture, which substantially reduces computational requirements and compensates for hardware gaps through algorithmic efficiency; ② Cluster breakthroughs, using clustered configurations to overcome single-card performance limitations and applying system-level thinking to redefine chip competition; ③ Advanced packaging and logic folding, as demonstrated by Huawei's Tao's Law, which restructures chip layouts through logic folding to circumvent lithography machine bottlenecks. Domestically produced GPUs will progress from functional to superior performance and eventually reshape the global chip landscape.

Notably, domestic GPUs are entering a period of commercial realization. On one hand, NVIDIA's market share in China's AI chip sector is shrinking rapidly, creating a historic window for domestic chip substitution. On the other hand, domestic manufacturers are pursuing differentiated pathways: through "mega-node" architectures that integrate vast numbers of chips into ten-thousand-card-scale clusters, achieving computational leaps at the system level. Domestic computing capacity is entering a golden period where demand expansion and market share gains reinforce each other.

The Huabao Shanghai Stock Exchange Innovation Board AI ETF (589520) and its linked funds (Linked Fund A: 024560; Linked Fund C: 024561) focus on China's domestic AI industry chain, holding 30 large-cap companies listed on the Shanghai Stock Exchange's innovation board that provide foundational resources, technology, and application support for AI. The semiconductor sector represents 70.4% of the fund's weighting, offering strong growth potential, with GPU-concept stocks and AI-application-concept stocks representing 41.98% and 23.19% of weights respectively. The 20% daily price limit makes the ETF an accessible tool for gaining exposure to breakthrough opportunities in the science and technology sector. Furthermore, the ETF is eligible for margin trading and securities lending, making it an efficient vehicle for positioning in domestic computing power.

*Institutional views referenced from economist Ren Zeping's July 9 report, "Semiconductor Industry Deep Dive: China's Domestic Substitution Wave in AI Computing Power Chips."

Source: Shanghai and Shenzhen stock exchanges, as of July 31, 2026. Reminder: Market volatility may be elevated in the near term; short-term price movements do not predict future performance. Investors are strongly advised to invest rationally according to their financial circumstances and risk tolerance, with careful attention to position sizing and risk management.

ETF fee disclosure: The ETF does not charge sales service fees. Securities transaction sponsors may charge commissions of no more than 0.5%, which include fees collected by securities exchanges, registration agencies, and other relevant institutions. On-exchange trading fees are determined by individual brokerage firms.

Risk disclosure: The Huabao Shanghai Stock Exchange Innovation Board AI ETF passively tracks the Shanghai Stock Exchange Innovation Board AI Index. The index base date is December 30, 2022; it was released on July 25, 2024. Index constituent stocks are adjusted periodically according to the index methodology; backtested historical performance does not predict future index performance. Individual stocks and index constituents mentioned in this article are presented for informational purposes only and do not constitute investment advice or represent any holdings or trading intentions of the fund manager. The fund manager assesses the Huabao Shanghai Stock Exchange Innovation Board AI ETF as carrying R4 (medium-high) risk level, suitable for aggressive (C4) and above-risk-tolerance investors; suitability matching opinions should be confirmed with your sales institution. All information appearing in this article—including but not limited to individual stocks, commentary, forecasts, charts, indicators, theories, and any form of statements—is for reference only, and investors are solely responsible for any self-directed investment decisions. The views, analyses, and forecasts herein do not constitute investment advice to any reader and bear no responsibility for direct or indirect losses arising from use of this content. Fund investment involves risk; past fund performance does not predict future performance; performance of other funds managed by the fund manager does not guarantee this fund's performance. Fund investment requires careful consideration.