Research
Loading latest report…
Loading research access…
Information on this site is for informational purposes only and is not investment advice. Always do your own research and consult a licensed advisor before making financial decisions. Data timestamps use US Eastern Time unless noted.
BofA Securities
Sep 17, 2026
Tags: China AI localization monetization semiconductors cloud AI applications strategy-view
Value Score: 9.0/10
Core Conclusion: BofA argues that China AI is moving from policy-led localization toward commercial monetization. The first phase was about closing gaps in chips, memory, equipment, and models; the next phase is about which parts of the stack can turn capability into durable profit. The report's key framework is a three-bucket equity map: China AI localization beneficiaries, global AI hardware/infrastructure exporters, and AI application names.
BofA separates China AI into three investable buckets with different risk-reward profiles. Localization plays benefit from policy support, domestic demand, and market-share gains, but they are often well-owned and richly valued. Global AI hardware exporters have proven revenue scale and global AI CAPEX leverage, but are highly exposed to the U.S. AI spending cycle and sanction risk. AI application names offer the largest long-term monetization potential, but near-term visibility is limited by competition, pricing pressure, and the absence of a clear killer app.
Chart takeaway: The framework shifts the discussion from a single "China AI" trade to three different equity problems: policy-backed substitution, global infrastructure cyclicality, and application monetization.
Linked conclusion: Portfolio construction should distinguish between localization beta, global AI CAPEX exposure, and application optionality.
China's AI localization is no longer only a policy ambition. In AI accelerators, BofA estimates domestic vendors captured about 50% of China market revenue in 2025, up from less than 30% in 2024, and could approach 80% by 2028E. In memory, CXMT's global DRAM revenue share rose from around 3%-4% in 2024 to about 10% by 2Q26, while YMTC's NAND share rose from 7%-8% to about 14%.
Chart takeaway: Domestic AI accelerator share is rising sharply, and Chinese memory vendors are gaining global share from a low base.
Linked conclusion: Localization is already showing up in revenue share, not just in policy language.
The progress is not uniform. Semiconductor equipment is a good example: domestic suppliers have made notable progress in etching, deposition, CMP, cleaning, and PR stripping, while lithography remains
