Goldman Sachs is urging clients to go long on China’s artificial intelligence value chain, arguing that global capital allocations have not caught up with the country’s growing role in AI-linked revenue, hardware production and infrastructure build-out.
In its latest strategy report, Goldman estimates that Chinese AI-related companies have a combined market capitalization of about $4 trillion and generate roughly 16% of global AI-related revenue. Yet global mutual funds hold only about 1.2% of their technology exposure in Chinese assets, leaving what the bank describes as a significant mismatch between economic contribution and market positioning.
The bank’s core argument is straightforward: China’s AI companies are already producing a meaningful share of global AI revenue, but global fund exposure remains extremely low. If capital flows begin to reflect China’s revenue share more closely, Goldman says the sector could see a broader repricing.
Since 2022, global AI-linked equities have added nearly $34 trillion in market value. Chinese companies account for close to 10% of that increase, according to the report, but they remain lightly represented in global technology portfolios. Goldman views that gap as the foundation of its long China AI trade.
The recommendation comes as traders reassess the next phase of the AI boom. While U.S. companies remain dominant in cloud computing, advanced graphics processors, large-scale AI platforms and commercial AI services, China is trying to strengthen the underlying industrial base that supports artificial intelligence: semiconductors, memory chips, data centers, power equipment, networking hardware and model development.
Goldman’s proposed basket is not centered on the consumer internet companies that drove earlier China technology rallies. Instead, it focuses on the physical and industrial layers of AI adoption, including electricity suppliers, chipmakers, AI infrastructure companies, model developers and application providers.
The bank said the theme is tied to China’s push for technological self-sufficiency and advanced computing capacity, both of which have become national priorities. That policy backdrop, combined with improving corporate earnings and expanding export demand, has made the AI value chain one of the most watched areas of China’s equity market.
Why Goldman sees a valuation gap
Goldman’s report frames the opportunity around a disconnect between revenue and ownership.
China’s AI-linked companies generate about one-sixth of global AI-related revenue, but global mutual funds allocate only a tiny portion of their technology exposure to Chinese assets. That means many global traders remain underweight China’s AI sector relative to its commercial footprint.
The bank argues that such a gap could narrow if fund managers become more willing to add exposure to Chinese technology shares, especially hardware and infrastructure names tied to national AI development. In that scenario, even modest changes in global allocations could have a meaningful effect because current positioning is so low.
Goldman estimates that AI-driven efficiency gains and new profit pools in China may still be underappreciated by the market. Its report suggests that the market may be pricing only part of the future earnings impact, with potential upside of 50% to 100% in selected areas if adoption, margins and policy execution improve.
The report does not suggest that China has overtaken the United States in AI. Instead, it argues that markets may be undervaluing China’s role in the global AI supply chain, particularly in hardware-heavy segments where state support and domestic demand are increasing.
This distinction is important. The U.S. AI trade has been led by cloud companies, advanced chip designers, software platforms and firms with clear revenue channels from AI services. China’s emerging AI trade is more industrial, with heavier emphasis on the infrastructure required to build and run AI systems at scale.
The trade moves beyond internet platforms
For much of the past decade, global exposure to Chinese technology was largely associated with internet platforms, e-commerce, gaming, digital advertising and online consumer services. Goldman’s China AI thesis is different.
The bank’s recommended basket spans electricity, semiconductors, AI infrastructure, model development and AI applications. That means the trade is less about consumer traffic and more about the capital-intensive build-out of computing capacity.
Data centers require power systems, cooling equipment, servers, memory, chips, networking gear and grid upgrades. AI model training requires huge amounts of computing power. AI deployment across factories, finance, health care, education, logistics and government systems requires additional hardware and software infrastructure.
Goldman believes this creates a broader opportunity across China’s industrial technology ecosystem. Rather than betting only on one or two platform companies, the trade captures suppliers that may benefit from rising AI adoption even if competition among app developers remains intense.
The report highlights hardware and infrastructure as areas receiving both policy attention and industrial funding. China’s leadership has repeatedly stressed the need to reduce reliance on foreign advanced technology, particularly in semiconductors and high-performance computing. That has pushed domestic chip development, memory production and computing infrastructure higher on the national agenda.
For traders, the significance is that AI is becoming linked not only to corporate spending but also to state-directed industrial planning. That could support demand through economic cycles, although it also increases dependence on policy delivery.
Exports point to stronger AI hardware demand
Trade data is one of the clearest signs that demand for China’s AI-related hardware is rising.
Chinese customs data showed exports increased 19.4% year on year in May. Within that figure, the value of integrated circuit exports rose by around 111%, even though export volumes grew only marginally. That combination suggests pricing power, stronger demand for higher-value products or improved product mix in areas tied to AI hardware and advanced electronics.
Earlier customs figures also showed integrated circuit export value climbing 83.7% year on year during the first four months of 2026. The sharp increase has supported the view that global demand for AI-related components is already feeding into China’s trade performance.
The official manufacturing purchasing managers’ index also expanded in June, with high-tech exports cited as one of the drivers. While a single month of data is not enough to confirm a durable trend, the improvement adds to evidence that China’s advanced manufacturing sector is gaining traction.
For AI-linked manufacturers, higher export values may also point to better capacity utilization. If factories are running at higher levels and selling more advanced components, earnings could improve faster than headline production data suggests.
Goldman’s report connects these export trends with the broader capital market thesis. If China’s AI hardware makers are generating more revenue and gaining share in selected product lines, the bank argues that global allocations may eventually need to adjust.
Beijing’s policy push is central
Policy support is another major part of the thesis.
Beijing is reportedly preparing a five-year program worth 2 trillion yuan, or roughly $295 billion, to build a nationwide AI data center network. Such a program would create demand across the AI infrastructure chain, including semiconductors, memory, servers, electrical systems, cooling equipment, cloud infrastructure and related construction services.
Reports indicate that the program may include requirements for domestic sourcing, with a large share of key technology expected to come from Chinese suppliers. If implemented, that would directly benefit local chipmakers, memory producers and data center equipment companies.
The scale of the potential program matters because AI infrastructure is expensive and energy intensive. Building nationwide computing capacity requires not only chips and servers but also land, power supply, cooling technology, fiber networks and grid reliability. That gives the policy push a wide industrial footprint.
China has already made technological self-reliance a central policy goal, especially after years of U.S. export restrictions affecting advanced chips and chipmaking tools. Those constraints have encouraged domestic firms to develop alternatives in semiconductors, memory, AI accelerators and software stacks.
Goldman’s view is that state support could lift companies across this chain, especially if policy funding turns into actual orders and revenue. Still, the bank notes that execution risk remains significant. Large-scale programs can face delays, cost overruns and uneven profitability across suppliers.
Market positioning has already started to shift
Capital markets have begun to react to China’s AI theme.
Recent index revisions across mainland China, Hong Kong and selected global benchmarks have increased the representation of AI and semiconductor sectors. Higher index weights can draw more passive flows into related stocks, especially when benchmark-tracking funds adjust their holdings.
China’s innovation-focused STAR Market 50 Index rose about 65% during the first half of 2026, reflecting stronger interest in domestic technology and semiconductor companies. The rally coincided with a broader reassessment of Asian AI exposure, as some traders looked beyond earlier regional favorites and toward China’s hardware supply chain.
Hong Kong-listed Chinese shares have also seen signs of renewed interest. The Hang Seng China Enterprises Index recently posted its strongest daily gain since early 2025, supported by optimism around technology policy, AI infrastructure and improving earnings prospects in selected sectors.
These moves do not eliminate risk. Chinese technology shares have experienced sharp rallies before, only to reverse when earnings failed to meet expectations or regulatory concerns resurfaced. But the current AI-linked rally is being supported by a different set of drivers: industrial policy, hardware demand, export growth and semiconductor localization.
Memory chipmakers show rapid growth
Corporate earnings from China’s memory sector are among the strongest examples of the momentum Goldman is highlighting.
Yangtze Memory Technologies reported a 445% year-on-year increase in revenue in the first quarter of 2026. The company’s global NAND flash market share rose from 8% to 13%, according to the report, and it is moving toward an initial public offering to finance expansion.
Changxin Memory Technologies, viewed as one of China’s leading domestic DRAM companies, recorded 50.8 billion yuan in first-quarter revenue. The company projected revenue of 110 billion to 120 billion yuan for the first half of the year.
Changxin’s IPO prospectus, disclosed on July 9, also forecast net profit for the first half of 2026 between 66 billion and 75 billion yuan. That would represent a year-on-year increase of more than 1,700%, positioning the firm as one of China’s most profitable technology companies for the period.
The growth reflects rising demand for memory products used in servers, AI infrastructure, smartphones, computers and industrial electronics. Memory is a critical part of the AI computing stack because AI workloads require massive data processing and storage capacity.
If Chinese memory companies continue to gain share, they could become key beneficiaries of both domestic AI infrastructure spending and global demand for advanced electronics. However, memory markets are cyclical, and profit margins can change quickly when supply expands or prices weaken.
The U.S. remains the benchmark
Despite Goldman’s positive stance on China’s AI value chain, the report acknowledges that the U.S. remains the global benchmark for artificial intelligence.
American companies lead in advanced GPU ecosystems, cloud infrastructure, large language model deployment, enterprise software integration and AI monetization. U.S. firms also have deeper links between chip design, cloud services and software platforms, creating mature revenue models that China is still trying to match.
This is one reason Goldman’s China thesis is built less on immediate AI software dominance and more on relative mispricing. The bank is not arguing that China’s AI ecosystem is risk-free or superior across all categories. It is arguing that China’s current market allocation appears too low compared with its revenue contribution and industrial momentum.
The biggest risks include policy execution, the profitability of state-backed infrastructure projects, continued U.S. technology restrictions, export controls, overcapacity and competition among domestic AI firms. Traders also must weigh currency movements, geopolitical tensions and broader sentiment toward Chinese equities.
Goldman’s report says the trade depends on follow-through. Policy announcements must translate into orders. Production increases must translate into margins. Export strength must prove durable. Semiconductor and memory companies must show that revenue growth can be sustained beyond a short-cycle rebound.
What could drive the next move
The next phase of the trade will likely depend on whether capital flows begin to close the gap Goldman identified: a $4 trillion Chinese AI-linked market value and a 16% share of global AI revenue, but only 1.2% exposure within global mutual funds’ technology allocations.
If global funds raise their China technology exposure, AI hardware and infrastructure names could see additional support. Index inclusion and higher benchmark weights may reinforce that move by increasing passive demand.
At the same time, the market will need confirmation from earnings, policy spending and operating performance. Traders will be watching whether data center construction accelerates, whether domestic chip requirements are enforced, whether memory producers keep gaining market share and whether AI applications generate measurable productivity gains.
Goldman’s call reflects a broader shift in how China’s technology sector is being viewed. The old China tech story was largely about consumer platforms. The new AI story is about computing capacity, chips, power, data centers and industrial self-reliance.
For now, the bank’s conclusion is that China’s AI value chain remains under-owned relative to its role in global AI revenue. Whether that gap closes will depend on how quickly policy support, corporate earnings and global capital flows move in the same direction.
Explore AI-driven trading potential beyond China’s infrastructure boom with Toobit’s smart AI copy trading strategies for volatile markets.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
