The Company Says Proven Yield and Stability Support AI, HBM4, And Automotive Chips
Exynos 2400 which adopts EXynos FinFET. (Photo courtesy of Samsung Electronics)
Samsung Electronics is accelerating its push to secure foundry customers by leveraging its 4-nanometer (nm) FinFET process, whose yield and stability have been validated through years of production. The move is widely interpreted as a strategy to broaden its order base by simultaneously targeting both leading-edge competition at the 2nm level and demand for mature process nodes.
On April 29, Samsung Electronics introduced the competitiveness of its 4nm FinFET process through its official semiconductor website’s technology blog. The company stated that the process has achieved a high level of maturity built on more than six years of mass production experience, and that accumulated production data has been used to enhance yield and process stability.
Samsung Electronics began mass production of its 4nm process in 2021. The 4nm process is the company’s fully mature process that delivers both performance and stability simultaneously. Cutting-edge processes offer superior performance but are initially difficult to control in terms of yield and mass production stability, while legacy processes are stable but limited in the degree of performance improvement they can deliver.
The company explains that its 4nm process has achieved both goals by occupying the space between these two extremes. It has accumulated approximately six years of sufficient mass production experience to ensure yield and stability, while also reaching a stage capable of meeting the latest performance requirements. A Samsung Electronics official said, “The 4nm is a process that is well-made and also delivers performance.”
In terms of performance, the company noted that it has improved interconnect resistance and capacitance (RC) delay by approximately 26% compared to the previous generation, thereby enhancing data processing efficiency. It also emphasized that support for a variety of threshold voltage (Vth) options enables customer-tailored design ranging from low-power products to high-performance semiconductors. The technology is assessed as being well aligned with the latest semiconductor market trend that demands both power efficiency and computational performance simultaneously.
Technologically, the 4nm process also stands out for the range of options it offers. Within the same process, it is possible to manufacture both performance-oriented and low-power products. For cases where performance matters, such as AI chips, speeds can be increased, while for applications where power management is critical, such as mobile and automotive, power consumption can be reduced through customized design. As the speed at which data moves within chips has increased, data movement efficiency has also improved.
Application areas are also expanding. Samsung Electronics noted that its 4nm process is well-suited for product categories requiring high performance and reliability, including artificial intelligence (AI) accelerators, base dies for sixth-generation High Bandwidth Memory (HBM4), automotive semiconductors, and radio frequency (RF) chips.
The company added that expanded application is also possible across a wide range of sectors, including mobile application processors (APs), chips for network equipment, and industrial system semiconductors. In addition, the process is capable of delivering stable performance even in power-constrained environments, making it increasingly useful in autonomous vehicles and similar applications. In the communications sector, it provides an environment well-suited for next-generation communication chips by enabling greater digital circuit integration while reducing power consumption.
Given that the 4nm process is sufficiently mature, production schedules and costs can be forecast to a considerable degree. This level of stability is a critically important factor in the semiconductor industry, where long-term supply assurance is essential. It is also one of the reasons the 4nm process is being adopted across diverse industries such as AI and high-performance computing (HPC), as securing yield becomes increasingly difficult as chip sizes grow larger, while the 4nm process provides a stable mass production foundation.
In particular, the 4nm process is playing an increasingly significant role in High Bandwidth Memory (HBM). Samsung Electronics is applying the 4nm process to the base die (logic die) at the lowest layer of HBM4, its sixth-generation HBM. Because HBM must process large volumes of data within a confined space, power and thermal management are central considerations, and the 4nm process is well-suited for such architecture as it can reduce power loss while increasing integration density. Samsung Electronics has been able to secure a competitive position in HBM4 through its 4nm process.
The industry is taking note of the fact that Samsung Electronics has re-highlighted its 4nm FinFET process alongside its next-generation 2nm gate-all-around (GAA) process. The move is widely analyzed as a strategy to absorb customer demand not only for cutting-edge processes but also for processes with proven production stability, thereby broadening the base of its foundry business. There are also assessments that, with the growth of the AI semiconductor market driving increased demand for large chip production, the importance of stable supply capability and yield assurance is only growing.
Samsung Electronics stated, “The 4nm FinFET process is equipped with the scalability to address a wide range of applications based on mature manufacturing competitiveness,” adding, "It is a platform capable of stably supporting the performance and efficiency that customers require.”
Kim Eun-jin
댓글
0