Investors holding Chinese stocks at a 38.5% discount to world markets get a forecast of tech leadership without returns, as price wars may keep China's AI industry cash-negative.
China is on course to lead the world in large language models and seven other strategic technologies by the end of 2030, and to be at least competitive in the remaining nine of 17 industries assessed, according to a report by Bloomberg Intelligence, Bloomberg’s research unit, published on October 28. The report was led by analyst Francis Chan, Ifeng reported.
The forecast does not crown either country overall in AI. The report expects the United States and China to share technology leadership in AI by 2030, with US developers possibly keeping a narrow edge at the frontier of model development. Its analysts wrote that neither country is likely to achieve overall supremacy, Bloomberg News reported.
The report’s sharper message for investors concerns returns. It argues China’s technical progress may outrun improvements in profitability, as overcapacity and price competition compress margins in AI, energy storage, hydrogen and robotics.
China already leads in robot production and parts of shipbuilding
The report says China already leads in humanoid robot production and in parts of shipbuilding, and that its research and development spending has caught up with that of the United States, measured on a purchasing-power basis, according to Taiwan’s TechNews. By 2030, it expects China to lead in large language models and optical transceivers, the components that carry data between servers in AI data centers, as well as in energy storage and biopharmaceutical manufacturing.
Its supporting figures, as reported by TechNews, show the scale behind those calls. China accounted for 28% of global manufacturing value added in 2024, more than the United States, Japan and Germany combined. Penetration of advanced driver assistance in China is forecast to rise from 11% in 2025 to more than 50% in 2030, and installed energy storage capacity to reach about 550 gigawatts by 2030. Major contract drug manufacturers held about US$38 billion in orders at the end of the first half, up 41% from 2023, helped by out-licensing deals.
On models, the report builds on earlier Bloomberg Intelligence work. In early October, senior analyst Robert Lea put the performance gap between leading US and Chinese models at a record low of 3% after DeepSeek’s V4.1 Flash release, HK01 reported, a finding EastFrontier covered when Bloomberg reported the narrower benchmark gap. Bloomberg Intelligence also credits Chinese developers’ skill in software optimization for positioning them well in lower-cost AI deployment.
Chips remain the constraint behind the AI forecast
Hardware is where the report sees China furthest behind. Chinese firms mostly trail today in AI hardware, from chips to optical networking components, but should be competitive by 2030, China Times reported. The report expects China to narrow the semiconductor gap without reaching full self-sufficiency.
The bottlenecks it names are advanced processors, high-bandwidth memory and chipmaking equipment. Domestic chip suppliers would meet only about one-fifth of China’s demand for cutting-edge AI training compute by 2030, Bloomberg Intelligence estimates. The domestic share of China’s semiconductor equipment market is projected to rise from 22% in 2025 to about 37% in 2030, according to TechNews.
That gap matters for the model forecast. Leading in large language models while relying on imported or constrained hardware for most frontier training implies Chinese developers will keep competing on efficiency rather than scale, which fits the report’s emphasis on low-cost deployment.
Profitability is the variable investors will track
Price competition among Chinese AI companies will delay profitability, and the sector may still be cash-negative by 2030 once capital spending is counted, according to TechNews. In his October report, Lea had said China’s AI industry might not turn profitable until 2030 and that sustainable profits would require cooling competition, consolidation and more rational pricing.
Markets already price that tension. MSCI China trades at 9.9 times forward earnings against 16.1 times for MSCI World, a discount of about 38.5%, the report noted. Evidence that would test its thesis includes margin trends at Chinese model developers in coming earnings, the pace of domestic equipment adoption, and whether Chinese suppliers win share in optical transceivers, the one AI hardware category where the report expects China to lead.