Photo: DIGITIMES
Samsung Electronics, one of the world's largest tech giants, is forecasting a sharp 56% decline in operating profits for the second quarter of 2025, underscoring the company’s ongoing challenges in adapting to surging demand in the AI chip sector. The South Korean tech giant said on Tuesday that it expects operating profit to come in at 4.6 trillion Korean won (approximately $3.36 billion), down from 10.44 trillion won a year earlier.
The forecast fell well below LSEG’s SmartEstimate of 6.26 trillion won, which is based on the most consistently accurate analyst forecasts. Samsung also projected revenue of 74 trillion won, missing expectations of 75.55 trillion won, a sign that both top- and bottom-line pressure is mounting despite broader industry momentum around artificial intelligence and semiconductors.
Shares of Samsung Electronics fell as much as 1.13% in early Tuesday trading following the release of the earnings guidance.
A key contributor to Samsung’s underperformance lies in its struggle to keep up with competitors in high-bandwidth memory (HBM)—a crucial component for AI processors. While Samsung has historically led in memory technology, rivals like SK Hynix and Micron have surged ahead in delivering cutting-edge HBM chips.
SK Hynix, in particular, has emerged as the preferred supplier to Nvidia, which controls roughly 70% of global HBM demand. Samsung, meanwhile, is still undergoing evaluation with Nvidia for its latest generation of HBM products. According to a local report, that certification process has been delayed until at least September.
Samsung declined to provide direct updates on its progress with Nvidia, stating only that its HBM products are "proceeding with customer evaluation and shipments." However, analysts say the delay significantly limits near-term growth potential.
Ray Wang, research director at Futurum Group, told CNBC: “Samsung has not yet passed Nvidia’s qualification process for its most advanced HBM, which significantly caps its upside this quarter. Even though they have secured supply agreements with AMD, those wins will likely not reflect in Q2 due to ramp-up timing.”
Samsung’s foundry business—its custom chip manufacturing arm—also remains a drag on performance. Orders have remained soft, and the division continues to face fierce competition from Taiwan Semiconductor Manufacturing Company (TSMC), the global leader in chip fabrication.
Additionally, U.S. export restrictions on advanced chips destined for China, one of Samsung’s major markets, have further squeezed demand. The company noted that inventory value adjustments and reduced sales of high-margin chips contributed to the profit decline.
In September 2024, Reuters reported that Samsung had directed some subsidiaries to cut staff by 30% in underperforming units, citing internal cost-reduction efforts across global divisions.
Despite the gloomy earnings forecast, Samsung Electronics' stock is still up more than 16% year-to-date, reflecting broader investor optimism about the long-term AI-driven semiconductor cycle. However, analysts remain cautious.
“The disappointing earnings are due to ongoing losses in the foundry business, while the upside in the high-margin HBM segment remains muted this quarter,” said MS Hwang, research director at Counterpoint Research.
Samsung is expected to release a more detailed breakdown of its second-quarter performance later this month. Market watchers will be looking for updates on HBM certification, competitive positioning with TSMC, and recovery signals from its memory and smartphone divisions.
Samsung’s second-quarter forecast reflects more than just a temporary dip—it highlights the challenges of navigating a rapidly shifting semiconductor landscape driven by artificial intelligence. With competitors capturing critical market share and geopolitical headwinds reshaping global chip supply chains, Samsung must move quickly to regain momentum. Investors, analysts, and partners alike will be watching closely to see whether the company can bridge the growing gap or risk falling further behind in the AI race.