Samsung, SK Q4 earnings outlook lowered amid weakening demand

The outlook for fourth-quarter earnings at Samsung Electronics and SK hynix is clouded by a negative market environment, driven by sluggish demand for information technology (IT) devices, and falling prices for legacy DRAM chips.

featured-image

Samsung Electronics' chip production line in Pyeongtaek, Gyeonggi Province / Courtesy of Samsung Electronics Broadcom's rise may provide new opportunity for chipmakers By Nam Hyun-woo The outlook for fourth-quarter earnings at Samsung Electronics and SK hynix is clouded by a negative market environment, driven by sluggish demand for information technology (IT) devices, and falling prices for legacy DRAM chips. According to market tracker FnGuide, brokerages estimate Samsung Electronics' operating profit for the fourth quarter at 9.22 trillion won ($6.

3 billion) as of Wednesday, a decrease of nearly 5 percent from 9.7 trillion won a month earlier. Throughout this month, local brokerages have been revising their estimates for Samsung’s operating profit, citing an oversupply of legacy memory chips due to sluggish demand for IT devices and undercutting by Chinese firms.



“In (the) fourth quarter, the memory market faces challenges stemming from excessive inventory in the chips for PC and mobile devices, necessitating further price cuts to boost shipping volumes,” iM Securities analyst Song Myung-sub said, lowering his estimation for the company’s fourth-quarter operating profit to 8.3 trillion won from 9.9 trillion won.

“Consequently, chances are high for Samsung's memory shipment volumes or average selling prices falling short of current market expectations.” According to DRAMeXchange, the average price of a DDR4 8GB 1Gx8 memory module, which is mostly used in PCs, dropped by 35.7 percent from $2.

10 in July to $1.35 in November. Market observers attribute the price drop to undercutting by Chinese memory manufacturers, with companies like CXMT and JHICC supplying DDR4 8GB chips at prices between $0.

75 and $1. The slowdown in the legacy DRAM business is impacting Samsung more significantly, as high-bandwidth memory (HBM) chips, which are in increasing demand for artificial intelligence (AI) processors, still account for only a small share of the company's earnings. “Samsung’s DRAM business is expected to underperform the market estimation because of delays in supplying HBM3e chips for Nvidia, CXMT’s undercutting and deteriorating supply and demand conditions for legacy DRAM chips,” Kiwoom Securities analyst Park Yoo-ak said.

SK Group Chairman Chey Tae-won, left, listens to an official at SK hynix's high-bandwidth memory chip production line in Icheon, Gyeonggi Province, Aug. 5. Courtesy of SK Group Although the situation is slightly better for SK hynix, which now stands as the world’s top supplier of HBM, the market is also lowering its estimation for the company’s fourth-quarter earnings due to the sluggish legacy chip market.

The market consensus for SK hynix’s fourth-quarter earnings was 8.05 trillion won as of Wednesday, down 1.2 percent from 8.

15 trillion won a month earlier. “Samsung Electronics and SK hynix will be able to rebound after showcasing sluggish performances in the fourth quarter of this year,” DAOL Investment & Securities analyst Ko Young-min said. The market expects that the two chipmakers will accelerate their transition efforts toward HBM to cope with the growing demand for AI processors.

Speculation is also rising that U.S.-based Broadcom is turning to the two Korean firms to secure HBM4, which will be produced starting next year, for its AI application-specific integrated circuits (ASIC).

The Korea Economic Daily reported earlier this week that SK hynix received Broadcom’s request for HBM4 and has begun developing the prototype. Samsung Electronics also initiated discussions with Broadcom over HBM4 supplies. Given that Nvidia has already secured most of the HBM chips that SK hynix will produce next year, industry officials expect that Samsung may cover orders from Broadcom.

While graphics processing units, or GPUs, like those used in Nvidia’s AI processors, offer more versatility for general computing, ASICs are more efficient for specialized tasks and are typically priced lower than GPUs..