The Shanghai Stock Exchange and China Securities Index Co. are launching a new benchmark for companies that build or deploy artificial intelligence applications, giving investors a defined way to follow part of China’s listed AI economy. The SSE STAR Market AI Applications Selected Index, code 950398, is scheduled for release on September 2. In its September 1 announcement, the exchange said the index is intended to reflect the performance of STAR Market securities involved in AI-application software and hardware.
The announcement is not the launch of a fund, nor is it a forecast of investment returns. It is an index methodology and a constituent list. That distinction matters because an index can influence how investors classify a sector and how financial products may later be built around it, but it does not guarantee new capital or commercial success for the companies included.
What makes the new benchmark useful is its attempt to define AI applications more broadly than foundation models. The methodology includes model development, industry-specific models, terminals and components that deploy or integrate those models, and software or technical services built on them. The resulting basket includes office software, cloud services, machine vision, robotics, cybersecurity, and data-related companies.
A 30-Stock Definition of AI Applications
According to the official methodology, the index will contain 30 stocks drawn from the STAR Market. Eligible securities must have been listed for more than six months, except for certain very large new listings, and must not be under delisting-risk treatment. They must also meet a liquidity screen that requires their average daily trading value over the previous year to rank in the top 90 percent of the eligible universe.
The selection process then excludes companies with a CSI ESG assessment of C or below. Remaining candidates must have businesses in three categories: basic and industry-specific model development; intelligent terminals, components, and integration technologies associated with those models; or software and technical services that apply model technology in particular business settings.
The final 30 are chosen by average daily total market capitalization over the prior year. This approach means the index is not a pure list of early-stage AI startups or a collection of whichever companies use “AI” most often in marketing. It favors liquid, larger STAR Market firms that fit the stated application criteria.
The official sample list includes Beijing Kingsoft Office Software, UCloud Technology, Roborock, Cloudwalk Technology, Shenzhen Intellifusion Technologies, Orbbec, Qi An Xin, and Beijing Haitian Ruisheng. Their inclusion demonstrates the cross-section the index is trying to capture. Some provide enterprise software or cloud infrastructure, some sell machine-vision or sensing capabilities, some work in robotics or connected devices, and others operate in security or AI data services.
Rules Matter More Than the Label
An index’s value depends on its rules. The new benchmark has a December 30, 2022 base date and a base value of 1,000 points. It is free-float market-cap weighted, with an individual constituent capped at a 10 percent weight. The cap is meant to prevent one large company from dominating the entire index, while still allowing market size to influence the basket.
Constituents will be reviewed twice a year, with adjustments implemented after the second Friday of June and December. The methodology also sets out procedures for temporary changes if a constituent is delisted or goes through events such as a merger, acquisition, or spinoff. These rules matter because they determine whether the index remains tied to its stated theme as the market changes.
The semiannual schedule creates a balance between stability and responsiveness. A benchmark that changes too often can become difficult to follow and expensive to replicate. One that changes too slowly can miss shifts in technology and market structure. The SSE and CSI have chosen a standard review cycle, leaving the details of company eligibility to the published framework rather than a discretionary narrative. That is a separate market mechanism from an individual company’s financing path, such as Enflame’s Shanghai IPO timetable for AI chips, but both developments make public-market access more relevant to the sector.
The index also illustrates how China’s public markets are trying to make AI investable through classifications. Recent listings and listing plans, including DapuStor’s effort to place AI storage in the market spotlight, show how the financing debate reaches well beyond foundation-model developers. The new index addresses a different part of the capital-market system. Instead of rules for companies seeking to list, it provides a benchmark for listed companies already involved in AI applications.
A Benchmark Is Not a Business Model
The new index can make it easier for investors to observe a group of AI-related companies, compare performance, and potentially create funds or derivatives in the future. But the announcement does not say that an exchange-traded fund will launch, that the group has a shared revenue profile, or that the companies will benefit equally from AI demand.
The constituents are exposed to different markets. An office-software company may have a different customer base and margin structure from a machine-vision provider. A robotics company faces manufacturing and deployment challenges that differ from cloud infrastructure. A cybersecurity business may have AI-related opportunities while still depending primarily on other sources of demand. The index is a classification tool, not a claim that every member has the same technology or financial outlook.
Investors should also avoid treating inclusion as an endorsement. The methodology selects companies according to eligibility, liquidity, ESG, theme, and market-cap rules. It does not establish that any particular firm has superior AI products, a guaranteed order book, or a safer investment profile.
Even so, the launch is a meaningful institutional signal. China’s AI policy has often been discussed in terms of labs, chips, and state plans. A dedicated STAR Market benchmark shows how the theme is moving into the everyday machinery of public markets, where analysts, asset managers, and retail investors use indices to organize attention.
EastFrontier’s coverage of Lightstandard’s 20 billion yuan valuation for optical AI compute illustrates why investors increasingly need ways to separate the many layers of the AI economy. The new index will not settle which companies will win, but it provides a transparent 30-stock map of how China’s leading science-and-technology board currently defines AI applications.