The remarkable surge of China's STAR 50 Index this year, which includes the nation's premier chip manufacturers, has been on par with the gains seen by the Philadelphia Semiconductor Index—home to AI giants like NVIDIA—and the exceptionally strong Korean KOSPI. The upcoming challenge for this market segment is to deliver stronger-than-expected earnings to justify its rapid ascent.
Spending on AI computing infrastructure by global hyperscale cloud providers, coupled with the Chinese government's enduring push to cultivate domestic technology champions, has propelled the chip-heavy STAR 50 Index to a record quarterly gain of 64%. This movement has been amplified by investors reallocating funds away from struggling consumer and retail stocks within China's A-share market.
While few doubt the long-term bullish narrative for China's tech supply chain, some investors are concerned that the recent sharp price increases and elevated bullish sentiment may have gotten ahead of themselves. As China's A-share earnings season commences this week with preliminary announcements, tech companies have an opportunity to validate the recent rally.
At the industry level, most analysts view this AI computing super-cycle in China's A-shares not as speculation on individual chips, but as a systemic diffusion across the AI computing supply chain, driven by the global frenzy for AI data center construction. Stronger demand for AI computing resources from both enterprises and consumers translates into higher power loads, more complex cooling requirements, faster optical interconnects/communications, higher-grade AI PCBs, and intense demand for advanced packaging capacity and cutting-edge upstream materials.
Recent projections from Wall Street institutions like Morgan Stanley and Goldman Sachs highlight that supply chain bottlenecks in AI computing infrastructure have expanded beyond "massive purchases of GPUs/ASICs" to encompass the entire AI data center delivery chain. This includes efforts to simultaneously address data center power equipment, liquid cooling, data center CPUs, DRAM/NAND/HBM, data center optical communications/interconnects, high-performance Ethernet network infrastructure, transformers, and gas turbines.
Furthermore, Morgan Stanley forecasts that nearly $3 trillion in AI-related infrastructure investment will flow through the global economy by 2028, with over 80% of this expenditure still ahead. The bullish thesis for Chinese stocks on Wall Street has shifted from simple valuation repair to a triple resonance of "robust AI tech profits + household asset reallocation + policy support."
Goldman Sachs notes that Chinese household allocations to equities remain low relative to long-term potential, currently constituting less than 10% of household assets. As the wealth effect from property diminishes and deposit rates stay low, household savings may gradually shift towards equity assets.
Following the STAR 50's 64% quarterly surge, China's hard-tech super-rally awaits validation during the earnings season. "Sentiment feels close to a temporary peak," remarked a senior fund manager from a Hangzhou-based asset management firm. "Many leading AI hardware players have already priced in the current strong growth, making it increasingly difficult to justify further gains based on fundamentals alone."
Funds chasing the AI computing theme in the A-share market have flowed into various niches of the supply chain, including key materials and components such as electronic cloth, copper foil, copper-clad laminates, glass substrates, and MLCC-related concept stocks. This level of spillover indicates that many investors lack confidence in continuing to chase the biggest winners in the US and Korean AI computing sectors, as well as the A-share AI computing giants, at current elevated levels.
Taking data center optical interconnect/communication equipment maker Zhongji Innolight as an example, its shares have skyrocketed 132% this quarter. Its forward P/E ratio now stands around 40x, compared to 33x for the Dow Jones US Telecommunications Equipment Index. The stock also trades about 20% above the average analyst price target compiled by institutions.
Broader earnings prospects have not entirely kept pace with share prices. Data compiled by institutions shows analysts have modestly raised overall profit expectations for the STAR 50 Index by just over 4% this quarter. This increase pales in comparison to the index's explosive rally, pushing its forward P/E to a new record high of 69x.
Nevertheless, most analysts maintain a bullish stance for now. Goldman Sachs prefers A-shares over Hong Kong-listed stocks, citing stronger profit performance for onshore investment indices driven by the shift of AI computing infrastructure-related profits to hardware companies. Morgan Stanley also favors A-shares due to their higher proportion of "hardcore" AI hardware tech firms.
This sharp divergence is partly reflected in benchmark index performance. While the STAR 50 soared, Hong Kong's Hang Seng TECH Index fell 5.3% this quarter, significantly lagging due to slower progress by major internet giants in developing AI application/agent ecosystems compared to their North American counterparts.
Some Chinese companies have already offered a glimpse of strong performance. Satellite Chemical, which produces chemicals for EVs and chips, indicated its first-half profit could surge up to 155%. Hangzhou Changchuan Technology, focused on semiconductor testing, stated its net profit would at least double. Shares of these companies rose sharply following the announcements.
The sector received another boost after China's top securities regulator pledged to ease listing standards for AI developers and other advanced tech firms, a move seen as a strong signal of support from Beijing, driving related stocks higher last week.
"Relative to global peers, domestic Chinese hardware tech stocks are expensive now and will likely remain so. But that alone is not a reason to turn bearish," said a fund manager at a Hong Kong-based asset management firm. "In this market, the playbook is simple: buy and hold. Don't overthink the comparison with global peers or get too hung up on valuation."
Another senior institutional investor believes that despite seemingly high valuations in some A-share sectors, China's tech upcycle is far from over. "The classic signals of a real downturn—weakening demand, policy tightening, and earnings misses—simply aren't present yet," they noted. "The next two years will be a critical window for breakthroughs in foundational technologies and key bottlenecks, with policy support clearly anchored until 2029."
Don't rush to exit Chinese tech stocks! High valuations meeting accelerating profits and household savings inflows could ignite the second half of the AI computing bull market. China's AI hardware rally has transitioned from "thematic re-rating" to "performance validation." While the STAR 50's record quarterly surge and the rapid valuation expansion of AI hardware leaders have led to signals like crowded positioning, concentrated trading, disappearing valuation discounts, and increased IPO supply—often seen near cycle peaks—global financial giant UBS still emphasizes "it's too early to exit." The core reason is that profits and orders have not yet weakened.
In UBS's latest China equity investment framework, expected profit growth for Chinese AI hardware tech stocks is around 80%, with earnings expectations revised up by an average of 15% over the past three months, while broader market earnings expectations were revised down 2% over the same period. Order visibility now extends to the end of 2027. UBS states these signs mean prices are expensive in the short term, but fundamentals have not yet entered a downturn. The real watershed is not "having risen too much," but when revenue growth, contract liabilities, inventory, and order visibility peak.
In Goldman Sachs' view, the global bull market around the AI computing chain is far from over. The market narrative has evolved from the long-standing post-2008 theme of "valuation expansion driven by programming/code-based software and light-asset models" to a "re-pricing centered on a series of physical assets for AI computing infrastructure."
At the specific AI computing industry chain level, Goldman Sachs believes the global AI bull market is far from declaring its end. It has moved from the first stage of an "AI chip purchasing frenzy" into a second stage of "massive construction of AI factories." The next round of excess alpha returns will not belong solely to the strongest leaders in AI GPU/ASIC fields but will systematically diffuse across the full stack of AI computing infrastructure for "AI factories." This includes data center high-performance CPUs, DRAM/NAND/HBM memory, AI PCBs, liquid cooling systems, data center optical interconnect systems, ABF substrates/glass substrates, MLCCs, electronic cloth, and a wide range of wafer foundry services.
NVIDIA CEO Jensen Huang stated last Wednesday that AI infrastructure could revitalize US factories, suggesting AI has the potential to usher in a new era of manufacturing and industrial growth in America.
Additionally, Goldman Sachs reiterates that Chinese household equity allocation remains low relative to long-term potential. With the property wealth effect waning and deposit rates low, household savings may gradually shift to equity assets. Therefore, Goldman Sachs maintains an overweight stance on the A-share market and has significantly raised its target for the CSI 300 Index to 5,500 points. By comparison, the CSI 300 hovered around 4,950 points on Friday.
Goldman Sachs's latest outlook provides a crucial underlying capital logic for the A-share AI hardware rally: even if foreign investors remain cautious about high valuations, the combination of low-yielding domestic deposits, a de-emphasis on property, and policy support for "hardcore AI tech" could continue channeling substantial local liquidity towards sectors like the computing chain, semiconductor equipment, power equipment, non-ferrous metals, and semiconductor raw materials.
China's AI computing bull market appears to be entering its most perilous, yet potentially most lucrative, later-stage primary surge. Valuations are no longer cheap; future gains can no longer rely on indiscriminate pushes from themes like "import substitution + grand AI narratives." They must be driven by real orders, entry into overseas supply chains, technological barriers, and profit cash flows, leading to performance stratification.
2027 is likely to be the real stress test year, when capacity expansion, a higher base, and intensified competition may cause high-beta, high-momentum small-cap stocks to gradually recede. However, segment leaders controlling AI computing bottlenecks—in areas like optical interconnects/modules, memory components, ASIC/GPU computing systems, copper-clad laminates, electronic cloth, PCBs, ABF/glass substrates, semiconductor equipment, and power infrastructure—may still emerge as the most enduring "hard AI tech主线" in China's asset revaluation. This is underpinned by the potential reallocation of household funds and policy support anchored until 2029.
Comments