There has been a clear realization within Samsung Electronics that things can't remain as they are because the semiconductor division is getting absolutely chewed up. Given that it generates the bulk of its revenues and profits, the company most certainly needs the division to regain its competitive advantage.
With this mind, Samsung recently made major changes to the leadership team, promoting EVP of its US chip business Han Jin-man to president and the man in charge of its entire foundry business. He has now shared an ambitious plan to tackle two of the company's most pressing threats: TSMC and China.
Samsung would be hoping it's not too little, too late
Taiwan-based TSMC is the world's leading contract chipmaker with a nearly 65% share of the market. Major companies like Qualcomm, NVIDIA, AMD, Apple, and others get their chips made on TSMC's advanced processors, including the current-generation 3nm process. TSMC is aiming to mass produce 2nm chips from next year to further extend its dominance.
Even though Samsung was the first to initiate the 3nm gate-all-around (GAA) process, commercialization has been way short of expectations, largely due to yield issues. That's why the foundry has been losing money hand over fist and has shut down some production lines because there aren't enough orders from customers.
On top of all that, there's the China threat, posed by the rapidly advancing China Semiconductor Manufacturing International Corp. OR SMIC. It has closed the market share gap to Samsung considerably, currently at 3.3 percentage points in Q3 2024 compared to 5.8 in Q2 2024. It would not be desirable for Samsung to be overtaken by SMIC.
During his inaugural address to foundry employees, Han Jin-man laid out Samsung's ambitious plans to tackle both of these challenges. He has a “two-track strategy” to close the technology gap with TSMC and keep SMIC at bay.
Han will focus on increasing yields of the next-generation 2nm process so that it can prevent all orders from going to TSMC. He'll also work on getting more customers for legacy processes of 10nm and above to ensure that all available foundry capacity is profitability utilized. Whether or not that's going to be enough is something that only time will tell.