As Tesla Chief Executive Elon Musk unveiled the recently finalized design of the AI chip "AI5" and expressed gratitude to foundry partner Samsung Electronics, the industry is focusing on the significance that Samsung Electronics' foundry division's 2-nanometer process has cleared its first hurdle after long-standing concerns.
The 2-nanometer process is the biggest battleground that will determine Samsung Electronics' foundry division's fortunes this year. However, the consensus is that it is too early to be relieved because Production yield and profitability have not yet reached maturity, and it is essential to secure additional clients beyond Tesla. In the past, even after Samsung Electronics became the first in the world to mass-produce the 3-nanometer process, it struggled to secure clients due to Production yield and performance issues.
According to the industry on the 17th, Samsung Electronics' 2-nanometer Production yield is in the mid-50% range, still falling short of around 60%, which is commonly cited as the threshold for stable mass production. Compared with competitor TSMC, which has already secured an 80%–90% Production yield, it is far behind. Samsung Electronics has begun mass-producing first-generation 2-nanometer products and plans to expand production of second-generation 2-nanometer products within the year.
Samsung Electronics' 2-nanometer process is not merely a miniaturization that narrows line width; it focuses on improving power efficiency. To that end, Samsung Electronics adopted the more advanced Gate-All-Around (GAA) architecture instead of conventional FinFET.
GAA is designed so that the gate surrounds the current-carrying channel on four sides, strengthening current control and reducing leakage current. It is particularly effective in advanced processes at 5 nanometers and below. And as fine processes evolve, it carries significant meaning in the AI Semiconductor era, where power efficiency and heat generation become more important.
But the most important factor is Production yield. In advanced processes, implementing the technology and producing it stably at scale are problems on entirely different levels. In fact, Samsung Electronics adopted GAA technology one to two years earlier than its competitors but has struggled in recent years to firmly establish the technology.
If Production yield does not reach a certain level, clients find it hard to have confidence in supply stability, and the more a foundry increases production volume, the greater its losses can grow. For Samsung Electronics' 2 nanometers as well, the key metric will likely be how quickly it can raise actual mass-production yield, rather than the technical completeness itself.
Securing profitability is also a tough challenge. Advanced processes require massive capital expenditures for facilities and research and development. On top of that, extreme ultraviolet (EUV) lithography equipment, advanced packaging, intellectual property (IP), and even back-end process integration all need to be upgraded. Ultimately, symbolic orders from one or two specific clients alone are unlikely to resolve the structural deficit in the foundry business.
A domestic foundry industry source said, "Profitability can improve only by continuously securing large clients and spreading fixed costs through a stable production system," and added, "Samsung Electronics must move beyond the level of a technology demonstration at 2 nanometers and build a truly 'money-making process' like TSMC."
The on-site responsiveness of the plant in Taylor, Texas, is also a key variable. Samsung Electronics aims to enhance responsiveness to big-tech clients including Tesla through an advanced foundry production base in the United States, but detailed "service strategies" such as bringing in cutting-edge equipment, securing client volumes, chip production yield, and meeting delivery deadlines are expected to affect future order volumes and client trust.