China’s semiconductor ambitions took another crucial step forward in 2025 with reports that Shanghai-based foundry Semiconductor Manufacturing International Corporation (SMIC) is testing domestically manufactured deep ultraviolet (DUV) lithography equipment developed by Sicarrier, an affiliate of Chinese optical giant SIAC. As first reported by TrendForce on September 17, 2025, this development signals a strategic inflection point for China’s AI chip production capabilities. The successful integration of DUV tools into advanced chip fabrication processes could allow Chinese foundries to bypass the international restrictions imposed by the U.S. and its allies on the export of advanced semiconductor equipment, particularly EUV (extreme ultraviolet) systems — a critical component for 7nm and smaller node production.
As global demand for AI chips continues to skyrocket, fueled by intense competition among large language models (LLMs), real-time generative AI applications, and AI in data centers, China’s push toward lithographic self-sufficiency becomes ever more significant. In this detailed exploration, we examine how the deployment of Chinese DUV Lithography tools may unlock the next phase of domestic AI chip development, the challenges it faces, and how it could impact the geopolitical and industrial AI landscapes in 2025 and beyond.
The Strategic Importance of Lithography in AI Chip Fabrication
Advanced lithography, particularly photolithography, plays a pivotal role in defining transistor features on semiconductor wafers. This makes the technology essential to achieving smaller process nodes, which directly affects chip performance, energy efficiency, and transistor density — all of which are indispensable in producing cutting-edge AI chips.
DUV lithography, though less advanced than EUV, can still be used to build nodes down to 7nm with multiple patterning techniques. While EUV systems, controlled primarily by Dutch firm ASML, dominate at the sub-7nm range, many high-performance AI chips — such as those used in edge devices and older cloud inference units — can still be economically produced using advanced DUV lithography paired with lithography multipatterning.
The ability for SMIC to deploy domestically produced DUV tools thus enables continuation of development at 14nm and potentially 7nm nodes, which are still viable for specific AI workloads. Further, China has shown proof-of-concept success with earlier SMIC 7nm chips — reportedly built via DUV multipatterning — as discovered through reverse engineering by TechInsights in mid-2022. This aligns with exploratory 2025 ambitions to expand AI performance per watt at lower cost using localized infrastructure.
Key Drivers Behind the Adoption of Chinese DUV Lithography
Regulatory Pressures and Sanctions on Semiconductor Equipment
Since 2019, growing export restrictions led by the U.S. and supported by Japan and the Netherlands have severely curtailed China’s access to cutting-edge EUV machines and, more recently, advanced DUV configurations like immersion DUV tools. These moves aim to delay China’s access to sub-7nm semiconductor manufacturing to preserve Western technological supremacy in AI and computing infrastructure. However, Sicarrier’s DUV solution — if proven commercially viable — represents a way for China to strategically work around these sanctions and satisfy urgent domestic demand for AI processing power.
AI Computation Growth in China
China is experiencing exponential AI computation growth, most notably in the training and deployment of large-scale models similar to GPT-4 and beyond. As reported by VentureBeat in August 2025, Chinese AI firms like Baidu, Alibaba, and Huawei have increased AI R&D budgets by over 18% YoY, focusing on foundational models and AI-as-a-platform services. These models require immense computational resources, underlining the importance of localized chip manufacturing to drive down costs and sustain scalability.
Cost-Effective Domestic Chip Supply Chains
Reducing dependence on foreign lithography not only mitigates geopolitical risk but also has financial benefits. Localized DUV manufacturing ensures domestic AI firms get priority access to hardware at reduced costs and shorter lead times, essential for faster AI model iteration cycles. According to McKinsey’s 2025 Semiconductor Rebalancing Report, countries capable of end-to-end fabs (design to packaging) see cost savings exceeding 25% per chip generation cycle over countries that must rely on staggered global supply chains.
Global AI Industry Implications and Potential Market Shift
China’s breakthroughs in domestic lithography will likely reshape key sectors of the global AI semiconductor industry. As the 2025 AI landscape becomes increasingly defined by generative AI, agent-based systems, and edge intelligence, control over chip design and manufacturing becomes a distinguishing factor. A functional DUV lithography ecosystem in China could have several implications:
- Increased Competition in Global AI Chips: More AI chip designs from China will surface in data centers and smart devices globally, intensifying price competition with NVIDIA, AMD, and Intel.
- Acceleration in AI R&D: Faster and cheaper access to AI hardware will enable Chinese developers to iterate and train models at a pace previously restricted by chip imports, potentially leading to new state-of-the-art models like Baidu’s “Wenxin5” or Alibaba’s “TanluVision.”
- National AI Infrastructure Growth: One of China’s national megaprojects — the “East-to-West Computing Resource Transfer Project” — plans to allow low-cost domestic chips to power inland data centers by 2026 to offset high power loads in coastal AI hubs.
All these developments could lead to a new global bifurcation in AI standards and silicon ecosystems, with Chinese models increasingly optimized on Chinese-made hardware running on AI chips fabricated with Chinese lithography tools.
Challenges and Limitations of Chinese DUV Technologies
Despite significant progress, it is essential to recognize the limitations China faces in fully replicating the sophistication of global lithography suppliers like ASML, Nikon, and Canon. DUV systems — even within immersion configurations — are complex pieces of machinery requiring ultraprecise optics, advanced photoresists, alignment systems, and power-laser engineering.
Sicarrier’s success depends on synergizing several industrial partners, including SMEE (Shanghai Micro Electronics Equipment) and NANC (National Analytical Center), which are still maturing their technology stacks. There is also the question of yield efficiency. While 7nm lithography can be achieved using DUV multipatterning, this method increases production cost, error rates, and wafer cycle time significantly compared to native EUV exposure. According to the 2025 SEMI whitepaper on lithographic yield, DUV-based multipatterning may reduce final yield rates by over 20% without advanced line-edge control algorithms and supporting materials from global suppliers currently restricted by export laws.
Moreover, without widespread adoption of high-numerical aperture optics, China’s DUV tools may struggle to keep pace with future demand for ultra-low power AI chips, such as those required for autonomous agents or foundation models operating on-device.
Comparative Landscape of Lithography Technologies (2025)
| Technology | Provider | Node Capability | Deployment Region | AI Relevance |
|---|---|---|---|---|
| EUV Lithography | ASML | 5nm → 3nm | US, Korea, Taiwan | Advanced training chips |
| Immersion DUV | Nikon, Canon | 7nm → 14nm | Japan, US, Taiwan | Data center inference |
| Chinese DUV (Sicarrier) | Sicarrier / SMIC | 14nm → 7nm* | China | Edge and domestic AI |
*Requires multipatterning and yields may vary
Conclusion: Geopolitical Stakes and AI Futures
As AI reshapes labor markets, governance, and security on a global scale, the ability to independently manufacture advanced chips — and the lithographic machinery that produces them — becomes a national imperative. The successful integration of Chinese-made DUV lithography tools into mainstream chip manufacturing suggests that China is inching closer to building an end-to-end AI chip ecosystem free from Western interventions.
This shift could accelerate the decline of a unified global chip standards model, creating dual hardware-software AI supply regimes: one Western-aligned and one China-centric. It may also spark innovation as emerging economies look to local alternatives — following China’s lead — to reduce costs and increase access to transformative AI technologies.