Samsung reduces memory chip output; shifts focus to advanced AI chips

Samsung Electronics, the world's largest memory chip maker, has announced plans to extend its production cuts of memory chips, including NAND flash chips used in smartphones and PCs, amid a prolonged slump in the global semiconductor market. However, the South Korean tech giant also said it would focus on producing high-end chips for artificial intelligence (AI) applications, which have shown strong demand and profitability.

Memory Chip Market Downturn

The memory chip market has suffered from severe oversupply and weak demand since late 2022 due to a global economic slowdown, trade tensions, and declining smartphone sales. Samsung's memory chip unit posted a record operating loss of 8.9 trillion won (US$7 billion) in the first half of 2023, compared to a profit of 19.9 trillion won in the same period last year.

To cope with the challenging market conditions, Samsung has been reducing its output of memory chips since April 2023, along with other major chip makers such as SK Hynix and Micron Technology. Samsung said that it would continue adjusting its production and inventory levels according to market demand and expects a gradual recovery in the second half of the year as customers deplete their chip stocks.

However, Samsung also warned that the demand recovery is largely limited to high-performance memory chips, such as high bandwidth memory (HBM) and low-power double data rate (LPDDR) chips used in AI, 5G, and cloud computing applications. These chips have higher margins and lower price sensitivity than conventional DRAM and NAND chips used in PCs and smartphones.

AI Chip Opportunity

Samsung sees AI as a key growth driver for its semiconductor business, as it aims to catch up with rivals such as Intel and Nvidia in the fast-growing market for AI chips. Samsung has invested heavily in developing its AI processors, such as the Exynos Auto V9 for autonomous driving and the Exynos 9820 for smartphones.

Samsung also said it would double its capacity of HBM chips by 2024, as it expects strong demand from major hyper scalers such as Google and Amazon, which use HBM chips to power their AI servers and data centers. HBM chips provide faster data processing and lower power consumption than conventional DRAM chips.

Samsung's strategy to shift its focus from memory chips to AI chips is expected to improve its profitability and competitiveness in the long term. However, Samsung also faces challenges such as technological complexity, regulatory uncertainty, and fierce competition from existing and emerging players in the AI chip market