SMIC reaches 5 nm N+3 volume production without EUV tools

SMIC has moved into volume production of its 5 nm-class N+3 node using only deep ultraviolet lithography, a full generation ahead of its earlier N+2 process, but yield and cost issues remain significant.

Chinese foundry SMIC has entered volume production of its latest 5 nm-class node, branded SMIC N+3, which is described as China’s most advanced semiconductor process manufactured without any extreme ultraviolet lithography tools. The company instead relies on deep ultraviolet equipment to produce its silicon, and a recent TechInsights teardown of the Huawei Kirin 9030 system-on-chip confirmed that this processor is fabricated on the new N+3 node. The move positions N+3 as a full generation ahead of the earlier SMIC N+2, which is a 7 nm-class node that Huawei has been using for its Ascend series of Artificial Intelligence accelerators and other infrastructure components.

The article explains that SMIC’s choice to continue with deep ultraviolet lithography introduces major technical hurdles as it pushes to smaller geometries. For advanced nodes, extreme ultraviolet scanners are said to provide more headroom because they operate at a wavelength of 13.5 nm, compared to the smallest deep ultraviolet immersion scanner wavelength of 193 nm. To compensate, SMIC is using aggressive deep ultraviolet multi-patterning techniques on N+3, which allows it to achieve 5 nm-class scaling despite the older lithography technology.

According to TechInsights, this deep ultraviolet-based approach to N+3 still comes with serious yield challenges, particularly around an aggressively scaled metal pitch on the new node. The analysis indicates that these manufacturing difficulties mean the Huawei Kirin 9030 system-on-chip is likely being produced at an operating loss, since a significant portion of dies must be discarded or reused for downgraded chips. While N+3 is presented as a step toward greater Chinese semiconductor independence at advanced process nodes, the report underscores that economic viability remains constrained by the technical limits of deep ultraviolet lithography at this scale.

55

Impact Score

OpenAI debuts GPT-5.2 on NVIDIA infrastructure for large scale training

OpenAI has launched GPT-5.2, describing it as its most capable model series yet for professional knowledge work, trained and deployed on NVIDIA’s full stack Artificial Intelligence infrastructure. The release underscores the growing importance of massive pretraining and post-training at scale using thousands of GPUs.

A decade of OpenAI and the road to superintelligence

Sam Altman reflects on OpenAI’s first ten years, from uncertain beginnings to systems that outperform top humans on difficult intellectual tasks, and lays out a confident path toward superintelligence over the next decade.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.