China AI Hardware Stocks Surge as Software Lags

Galaxy Securities, one of China's largest brokerage firms, has issued a strategic recommendation urging investors to prioritize upstream AI computing infrastructure stocks, citing strong earnings performance and a widening divergence between hardware and software segments in the country's artificial intelligence sector. The April 2025 analysis highlights a structural split that mirrors — yet differs from — trends seen in Western AI markets.

The call comes as China's domestic computing power supply chain delivers earnings that 'significantly exceeded expectations,' drawing capital toward companies with strong performance certainty and high growth forecasts.

Key Takeaways at a Glance

• Hardware dominance: China's AI sector shows a stark hardware-strong, software-weak structure in April 2025
• Earnings surprise: Upstream computing infrastructure companies, especially domestic chipmakers, posted results far above analyst expectations
• Geopolitical fade: The impact of geopolitical tensions and elevated oil prices on AI stocks has weakened in the short term
• Capital rotation: Investors are shifting toward sectors with high earnings certainty and strong forward guidance
• Structural divergence: The gap between hardware and software/application performance continues to widen
• Top pick: Galaxy Securities specifically recommends China's domestic computing power supply chain as the highest-conviction play

Hardware Outperforms as Software Struggles to Monetize

The divergence within China's AI sector tells a familiar story with a uniquely Chinese twist. While companies building the physical infrastructure for AI — chips, servers, networking equipment, and data center components — are posting blockbuster quarterly results, software and application-layer companies are struggling to demonstrate comparable revenue growth.

This pattern echoes what Western markets experienced throughout 2023 and 2024, when Nvidia surged over 200% while many AI application startups struggled to convert user growth into sustainable revenue. However, China's version of this divergence carries additional complexity due to U.S. export restrictions on advanced semiconductors, which have accelerated domestic chip development efforts.

Galaxy Securities notes that the structural split is not merely a temporary phenomenon. The brokerage expects the hardware-software gap to persist in the near term, driven by the fundamental reality that AI infrastructure must be built before applications can scale. Capital is flowing toward companies with proven revenue streams rather than speculative application-layer bets.

Why Domestic Computing Power Is China's Hottest AI Trade

The emphasis on domestic computing power (国产算力) reflects a strategic imperative unique to China's AI ecosystem. With Washington maintaining tight restrictions on exports of advanced AI chips — including Nvidia's H100 and A100 GPUs — Chinese companies have been forced to develop indigenous alternatives at an unprecedented pace.

This constraint has paradoxically created a massive investment opportunity. Companies like Huawei (with its Ascend AI chip series), Cambricon Technologies, and Hygon Information Technology have seen surging demand as Chinese cloud providers, tech giants, and government agencies seek domestically produced computing solutions.

The earnings season has validated this thesis. Several key players in China's domestic computing supply chain reported revenue growth exceeding 50% year-over-year, with some component suppliers seeing even more dramatic increases. The results have attracted institutional capital that previously sat on the sidelines, waiting for fundamental confirmation of the domestic chip substitution narrative.

Key companies benefiting from this trend include:

• Huawei Ascend ecosystem partners and suppliers
• Cambricon Technologies — China's leading AI chip designer
• Hygon Information Technology — x86-compatible server processors
• Montage Technology — memory interface and server chips
• Inspur Electronic Information — AI server manufacturing

Geopolitical Headwinds Fade, Earnings Take Center Stage

One notable shift highlighted in Galaxy Securities' analysis is the diminishing impact of geopolitical tensions and elevated oil prices on AI sector performance. Earlier in 2025, concerns about potential escalation in U.S.-China tech restrictions and energy cost inflation weighed heavily on sentiment across the sector.

However, as earnings season progressed, strong fundamental results have overshadowed macro concerns. Investors are increasingly making decisions based on company-specific performance metrics rather than broad geopolitical narratives. This shift suggests a maturing market where AI infrastructure investments are evaluated on their own merits.

The reduced sensitivity to oil prices is particularly noteworthy given the energy-intensive nature of AI data centers. China's aggressive buildout of renewable energy capacity and its relatively lower electricity costs in key data center regions — such as Guizhou, Inner Mongolia, and Ningxia — have helped insulate computing infrastructure companies from energy price volatility.

Compared to the U.S. market, where data center energy consumption has become a headline concern — with companies like Microsoft and Amazon signing nuclear power agreements — China's approach of locating facilities near cheap hydroelectric and wind power has proven cost-effective.

The Global AI Infrastructure Boom in Context

China's computing infrastructure surge does not exist in isolation. Globally, AI infrastructure spending is projected to exceed $300 billion in 2025, according to estimates from multiple research firms. The race to build computing capacity is arguably the defining investment theme of the decade.

In the United States, Nvidia continues to dominate the GPU market with its Blackwell architecture, while AMD gains ground with its MI300X accelerators. Intel is attempting a comeback with its Gaudi 3 AI accelerator. Cloud hyperscalers — Microsoft Azure, Amazon Web Services, and Google Cloud — are simultaneously developing custom silicon to reduce dependence on third-party chip suppliers.

China's parallel infrastructure buildout creates a fascinating dual-track dynamic. While Western companies operate with access to the most advanced semiconductor manufacturing (primarily through TSMC in Taiwan), Chinese firms are innovating within tighter technological constraints. This has led to creative architectural solutions, including chiplet-based designs that combine multiple smaller chips to approximate the performance of monolithic advanced processors.

The investment implications extend beyond China's borders. Western companies supplying non-restricted components — such as certain types of memory chips, passive components, and data center cooling systems — benefit from demand on both sides of the geopolitical divide.

What This Means for Investors and Industry Watchers

Galaxy Securities' recommendation carries several practical implications for different stakeholders:

For global investors:
- China's AI hardware sector offers exposure to a structural growth trend driven by both organic demand and import substitution
- The earnings confirmation reduces the speculative nature of these investments compared to 12 months ago
- Currency and regulatory risks remain relevant considerations for foreign investors accessing these stocks through Hong Kong or mainland exchanges

For AI industry participants:
- The hardware-software divergence suggests that monetization challenges in AI applications are a global phenomenon, not unique to any single market
- Infrastructure buildout timelines indicate that the 'picks and shovels' trade still has significant Runway
- Domestic supply chain development in China could eventually create new competitors for Western chip companies in emerging markets

For technology strategists:
- The parallel development of computing ecosystems — one U.S.-led, one China-led — is accelerating, creating potential interoperability challenges
- Companies operating across both ecosystems need dual-sourcing strategies for critical AI infrastructure components

Looking Ahead: How Long Can Hardware Lead?

The critical question for investors following Galaxy Securities' advice is whether the hardware-software divergence can persist. Historical patterns from the internet era suggest that infrastructure buildout phases eventually give way to application-layer value creation — but the timeline can be longer than most investors expect.

In the near term, several catalysts could sustain hardware momentum. China's government continues to prioritize AI infrastructure in its industrial policy, with provincial governments competing to build computing centers. The upcoming release of next-generation domestic AI chips in the second half of 2025 could provide additional catalysts for the supply chain.

However, risks remain. A significant escalation in U.S. export restrictions could disrupt supply chains for critical components that Chinese manufacturers still source internationally. Additionally, if AI application revenue fails to materialize at scale, the economic justification for continued infrastructure investment could weaken.

For now, the market's message is clear: in China's AI sector, the money follows the hardware. Galaxy Securities' recommendation aligns with this consensus, but adds the important nuance that domestic computing power — not just any hardware play — represents the highest-conviction opportunity. As the AI infrastructure race intensifies globally, the performance gap between nations building their own computing foundations and those dependent on imports will only grow wider.