The AI industry chain experiences low-level fluctuations and divergence, with communication and IT services remaining active, while the chip sector leads the decline.

Ask AI · What technical iteration factors lie behind the differentiation in the AI industry chain?

On March 30, the overall AI industry chain showed a low-level, volatile pattern. Within the sectors, performance diverged: communication equipment and IT services were relatively active, while the chip design sector led the declines, dragging down related indices. By the close, the CSI Artificial Intelligence theme index fell 0.4%, the CSI STAR Market and ChiNext Artificial Intelligence index fell 0.5%, and the Shanghai STAR Market Artificial Intelligence index had the largest drop, at 1.1%.

Judging from index characteristics and valuation (data source: Wind, as of March 27, 2026):

CSI Artificial Intelligence theme index: trailing price-to-sales ratio of 4.7x; since its launch in 2015, the valuation percentile has been 90.1%; the combined share of the computer, electronics, and communications industries is about 90%, covering leading A-share AI full-industry-chain companies;

CSI STAR Market and ChiNext Artificial Intelligence index: trailing price-to-sales ratio of 13.7x; launched in May 2025, focusing on the core AI of the STAR Market and ChiNext; the shares of the electronics, communications, and computer industries are nearly 90%;

Shanghai STAR Market Artificial Intelligence index: trailing price-to-sales ratio of 12.8x; launched in July 2024, focusing on the STAR Market’s compute power and application layers; the combined share of the electronics and computer industries exceeds 85%.

See the table below for the details of the Artificial Intelligence series ETFs under E Fund:

China Academy of Information and Communications Technology (CAICT) points out that, against the backdrop of accelerated iteration of AI technologies, China’s demand for intelligent computing power is shifting from large-scale expansion to more efficient upgrades. AI computing nodes have become the core unit supporting the development of intelligent computing power. This architecture integrates multiple computing-power chips through high-speed interconnect, effectively addressing the challenges of compute co-operation and efficiency in large-model training, injecting new momentum into industrial development.

Risk warning: Funds involve risk; invest with caution.