Samsung Electronics Co.'s profits fell for the first time since 2023. This was hurt by a US curb of Chinese-bound AI chips and hiccups in a plan to sell cutting-edge memory to Nvidia Corp.
South Korea's largest company reported a reserve operating profit of 4.6 trillion won in the June quarter, down about 56% from a year ago. Analysts had an average decline of 41%. The revenue was 74 trillion won.
According to Samsung, one-time inventory-related costs have contributed to lowering, with customer evaluations and shipping of advanced memory products ongoing. Samsung said operating losses in the contracted chipmaking business are expected to narrow in the second half of the year, as demand gradually recovers.
The company will provide full financial statements later this month, including net income and divisional breakdowns.
Samsung has struggled to regain its footing with the high-bandwidth memory chip, which is essential to powering Nvidia's AI accelerators. The company has yet to secure certification from NVIDIA for its most advanced product, the 12-layer HBM3E. Meanwhile, US competitor Micron Technology Inc. is moving forward rapidly and making its own claims.
Analysts voted by Bloomberg News prior to the reserve revenue release found that Samsung's chip division recorded operating profit of 2.7 trillion won in the second quarter, earning 2.7 trillion won from the previous quarter, significantly lower than the previous year's 6.5 trillion won.
In April, Samsung signaled a better outlook, shipping enhancements to its HBM3E samples to key customers, predicting that its product line would contribute to revenue in the second quarter. The company also said it plans to start mass production of HBM4 chips later this year.
Samsung is fighting to catch up with SK Hynix, which is actively positioned as Nvidia's leading HBM4 supplier. The world's first 12-layer HBM4 sample was shipped to customers earlier than expected, followed by microns in June, but Samsung had to modify the design of the 12-layer HBM3E.
According to the June release, Samsung secured orders from Advanced Micro Devices Inc. and joined Micron as a supplier. However, it failed to earn early certification for the HBM3E chip from Nvidia, the dominant manufacturer of AI-backed graphics processing units – has hurt attempts to take away important market share.
Analysts at Bernstein, led by Mark Lee, who previously expected Samsung's 12th tier HBM3E to be eligible by NVIDIA in the second quarter, said they trimmed Samsung's forecast of HBM market share and are hoping to be certified in the third quarter.
“Samsung will gradually narrow down the gaps and rivals,” they wrote in their June 23rd research notes. “We expect SK Hynix to be the leader in 2027, but as others catch up and erode the edge of Sk Hynix, the stock held by suppliers will be distributed more than it is now.”
Bernstein estimates that SK Hynix holds 57% of the HBM market in 2025, with 27% and Micron at 16%.
At its annual shareholders meeting in March, Samsung vowed to strengthen its position in the HBM market this year, addressing concerns about its poor performance in AI. Jun Young Hyun, head of Samsung's chip business, said Samsung's failure to secure an early lead in the HBM market contributed to its lag behind rival SK Hynix and promised not to repeat mistakes on the HBM4. The next generation of memory will be used in Nvidia's Rubin GPU architecture.
“Uncertainty remains around the timeline for quality certification, but no signs of a strategic shift are seen and the company believes it is on track for the launch of the 3Q2025 market,” Daishin Securities Co. Analyst Ryu Hyung-Keun wrote in a recent research report.
With the support of Shinhye Kang.
This article was generated from an automated news agency feed without any text changes.
