Journalist

김혜준
Candice Kim
  • Samsung Electronics eyes quarterly operating profit of 30 trillion won
    Samsung Electronics eyes quarterly operating profit of 30 trillion won SEOUL, February 16 (AJP) - Samsung Electronics is expected to reach 30 trillion won in quarterly operating profit for the first time in the January-March period, driven by surging demand for AI-related memory chips. The South Korean tech giant posted a record 20 trillion won in operating profit in the October-December quarter, becoming the first domestic company to reach that milestone. Analysts now project it will add another 10 trillion won in just one quarter. According to brokerage estimates, Samsung's first-quarter operating profit is forecast at 32.5 trillion won, up 386.6 percent from 6.7 trillion won a year earlier. Revenue is expected to reach 111.4 trillion won, a 40.8 percent increase from 79.1 trillion won in the same period last year. Rival SK Hynix is also projected to approach the 30 trillion won mark, with analysts estimating first-quarter operating profit of 28.3 trillion won, up 280.2 percent year-on-year. The company's sales are forecast at 42.9 trillion won, a 143.1 percent jump from the previous year. SK Hynix posted an operating profit of 19.2 trillion won in the fourth quarter, just short of the 20 trillion won threshold. However, analysts say the company could reach 30 trillion won in the current quarter, with its operating margin expected to hit 66 percent, exceeding the record 58 percent achieved in the previous quarter. The exceptional performance stems from explosive demand for high-bandwidth memory and DRAM chips used in AI servers, as well as NAND flash memory for data centers. Memory chip prices have surged as hyperscalers rush to secure supply for AI infrastructure buildouts. Market research firm Counterpoint Research projects memory prices will rise 80-90 percent in the first quarter compared to the previous quarter. Prices for 8GB DDR4, a standard PC memory chip, are expected to jump 91 percent after a 35 percent increase in the fourth quarter. Server-grade 64GB DDR5 prices are forecast to surge 99 percent following a 76 percent gain in the previous quarter. Global investment bank Morgan Stanley estimates Samsung's operating profit will reach 245.7 trillion won this year and 317.4 trillion won next year. For SK Hynix, the bank projects 179.4 trillion won this year and 225.4 trillion won in 2027. Nomura Securities raised its target price for Samsung to 290,000 won from 220,000 won, and for SK Hynix to 1.56 million won from 1.25 million won. The brokerage projected Samsung's operating profit at 243 trillion won this year and 322 trillion won next year. KB Securities analyst Kim Dong-won noted that memory supply shortages have intensified in February compared to the fourth quarter, with major customers' demand fulfillment rates at around 60 percent. "AI data center companies are absorbing 70 percent of Samsung's memory shipments, establishing a solid structural demand base," Kim said in a recent report on Thursday. "Considering the limited expansion of memory supply through 2027, maximum production capacity will emerge as a core competitive advantage." Industry sources say both companies have already sold out their memory production for this year, with global big tech firms lining up with advance payments to secure supply, providing unprecedented earnings visibility and stability compared to previous boom cycles. 2026-02-16 14:29:16
  • South Korea adds speed skating top-10 finish, falls short in skeleton mixed team debut
    South Korea adds speed skating top-10 finish, falls short in skeleton mixed team debut SEOUL, February 16 (AJP) - South Korea's Lee Na-hyun placed 10th in the women's 500-meter speed skating on Sunday, marking her second top-10 finish at the Milan-Cortina Winter Olympics, while the country's skeleton mixed team finished 11th in the inaugural Olympic event. Lee clocked 37.86 seconds at Milano Speed Skating Stadium, finishing 0.59 seconds behind bronze medalist Miho Takagi of Japan. The 20-year-old had earlier placed ninth in the women's 1,000 meters, making her the first South Korean to crack the top 10 in that event. Starting from the inside lane in the 13th pairing, Lee passed the first 100 meters in 10.47 seconds, the eighth-fastest opening split. She navigated the first curve smoothly and powered through the straightaway but lost some speed on the final curve, unable to overcome centrifugal force. Kim Min-sun, competing in her third Olympics, finished 14th with a time of 38.01 seconds. The 30-year-old struggled from the start, passing the 100-meter mark in 10.61 seconds, and was unable to close the gap despite her trademark finishing speed. Netherlands' Femke Kok claimed gold with an Olympic record of 36.49 seconds, 0.66 seconds ahead of compatriot Jutta Leerdam, who took silver. Takagi earned bronze with 37.27 seconds. In skeleton, Jeong Seung-gi and Hong Su-jung combined for 2 minutes, 1.45 seconds to finish 11th out of 15 teams in the mixed team event at Cortina Sliding Centre. The format, making its Olympic debut at Milan-Cortina, features one woman and one man from each team completing runs with their times combined. Hong, competing in her first Olympics, recorded 1:01.84 despite posting the fastest reaction time of 0.13 seconds. She struggled with control early in her run, reaching a top speed of 120.48 kilometers per hour before finishing 13th among women. Jeong, who also placed 10th in the men's individual event earlier in the Games, posted 59.61 seconds for 11th place in the men's portion. The 27-year-old showed strong technical skills with a reaction time of 0.19 seconds, sixth-best overall. Britain's Matt Weston and Tabitha Stoecker won the inaugural mixed team gold with a track record of 1:59.36. German teams swept silver and bronze, with Axel Jungk and Susanne Kreher taking silver in 1:59.53, and Christopher Grotheer and Jacqueline Pfeifer claiming bronze in 1:59.54. In women's curling, South Korea defeated Japan 7-5 in their fifth round-robin match at Cortina Curling Olympic Stadium. Skip Kim Eun-ji's team improved to 3-2 after earlier losses to the United States and Denmark. Trailing 3-3 entering the eighth end, South Korea scored three points for a big end. The team then held a one-point lead heading into the 10th end with the hammer, adding one more to secure victory. South Korea currently sits tied for fourth place with Denmark in the 10-team round robin. The top four teams advance to the semifinals, with the medal rounds scheduled for Feb. 20-22. 2026-02-16 11:35:10
  • SK chair meets US big tech leaders for HMB4 pitch
    SK chair meets US big tech leaders for HMB4 pitch SEOUL, February 13 (AJP) -SK Group Chairman Chey Tae-won commanding the world's most-sought memory chip unit held a string of meetings with big-tech leaders in the United States to strengthen leadership in AI chip supply chain amid tension with Samsung Electronics over HBM4 supply. His business meetings included all the big tech names Nvidia, Broadcom, Microsoft, Meta and Google engaged in AI chip accelerators, as he pushed to expand partnerships centered on high-bandwidth memory (HBM) and next-generation AI infrastructure. Chey first met Jensen Huang, chief executive of Nvidia, in Santa Clara on Feb. 5 (local time). The two exchanged views on global AI industry trends, demand for advanced semiconductors and cooperation on next-generation GPUs and HBM. Ahead of the meeting, Chey reviewed SK hynix’s supply strategy and shifts in the AI ecosystem, the company said. On the following day, Chey visited Broadcom’s headquarters in San Jose for talks with CEO Hock Tan. Their discussions focused on the medium- to long-term outlook for the memory market, supply strategies and investment cooperation. The two sides also explored ways to reflect SK hynix’s HBM technology in Broadcom products from the design stage, as Broadcom develops customized AI accelerators and networking solutions. On Feb. 10, Chey met Satya Nadella in Seattle to discuss expanding cooperation beyond semiconductors to include AI data centers and cloud-based solutions, reflecting SK Group’s broader push into AI infrastructure and services. The same day, Chey held talks with Meta CEO Mark Zuckerberg in San Jose on long-term memory supply and next-generation AI infrastructure. SK hynix shared its HBM strategy optimized for Meta’s in-house AI accelerator project, MTIA, and discussed early alignment of technology road maps for products beyond HBM4 and HBM4E. On the final day of his trip, Feb. 11, Chey met Sundar Pichai at Google’s campus in San Jose. The two agreed that securing memory supply has become a key bottleneck in expanding AI data centers and that stabilizing medium- to long-term supply is critical, as production capacity cannot be expanded quickly. They also discussed cooperation on custom HBM and future HBM4-based chips aligned with Google’s AI roadmap. SK hynix said the meetings underscore its efforts to deepen partnerships with global tech companies based on its AI memory leadership. The company aims to expand HBM supply while building cooperation that spans next-generation chips, data centers and cloud infrastructure. 2026-02-13 16:42:43
  • CJ Olive Young unveils wellness pop-up at Seongsu flagship store
    CJ Olive Young unveils wellness pop-up at Seongsu flagship store SEOUL, February 13 (AJP) - CJ Olive Young has launched a pop-up store dedicated to its new wellness curation platform at its flagship location in Seongsu-dong, a district that has cemented its status as Seoul’s primary retail testbed. The country’s top beauty and health retailer announced on Friday that it opened the "New Wellness Lounge" on the first floor of "Olive Young N Seongsu." The pop-up, which runs until March 9, is designed to introduce the lifestyle concepts proposed by its newly launched content platform, "Olive Better." The launch comes as Seongsu-dong continues to serve as a barometer for consumer trends in South Korea, with major retailers increasingly utilizing the district to pilot new branding strategies and engage with trend-conscious demographics. The "New Wellness Lounge" focuses on three core themes: "Eat Well," "Fit Well," and "Relax Well." Visitors can experience a curated wellness routine that includes healthy snacks, stretching guides, and aromatherapy zones. The space also features interactive elements such as a "check-in" event linked to social media, offering visitors a hands-on experience of the company's wellness proposition. "The lounge is meaningful as a space created together by Olive Young N Seongsu, a trend hub, and Olive Better, a wellness curation platform," an Olive Young official said. "We will continue to plan various contents to help customers discover their own wellness routines." In parallel with the Seongsu pop-up, CJ Olive Young is operating an "Olive Better Lucky Board" event near Hongik University Station in western Seoul until Feb. 18, offering gift cards and promotional products to engage a wider audience. 2026-02-13 14:43:39
  • Samsung Electronics starts mass shipments of HBM4, targets next-gen AI demand
    Samsung Electronics starts mass shipments of HBM4, targets next-gen AI demand SEOUL, February 12 (AJP) - Samsung Electronics has begun mass shipments of the industry's first HBM4 high-bandwidth memory chips, the South Korean tech giant said on Thursday, looking to secure an early lead in the next-generation AI memory market. The new chip utilizes Samsung's advanced 10-nanometer class (1c) DRAM and its proprietary 4-nanometer foundry process for the base die. It delivers stable transfer speeds of 11.7 gigabits per second (Gbps) and hits a maximum of 13 Gbps, significantly exceeding the industry standard of 8 Gbps set by JEDEC. Samsung said the HBM4 offers a peak bandwidth of up to 3.3 terabytes per second (TB/s) per stack. It supports capacities ranging from 24 gigabytes (GB) to 48 GB, depending on the stacking configuration (8, 12, or 16-high). The company highlighted that its integrated manufacturing capabilities—covering memory, foundry, and packaging—allowed it to optimize power delivery. The new chips improve energy efficiency by approximately 40 percent and reduce thermal resistance by 10 percent compared to previous generations, key factors for reducing operating costs in data centers. "Samsung’s HBM4 departs from conventional methods by adopting cutting-edge technologies like 1c DRAM and a 4-nanometer foundry process right from the start," said Hwang Sang-joon, Executive Vice President of Memory Product & Technology. "We have secured sufficient headroom for performance scaling to meet customers’ rising demands in a timely manner." Looking ahead, Samsung plans to ship samples of the upgraded HBM4E in the second half of 2026 and begin sampling custom HBM products in 2027. Buoyed by the expansion, the world's largest memory chipmaker expects its HBM revenue in 2026 to more than triple from 2025 levels. 2026-02-12 16:36:12
  • Seongsu – how pop-up fever turned shop space into show space
    Seongsu – how pop-up fever turned shop space into show space Editor’s Note: This is the second installment in AJP’s Seongsu series, which examines how Seoul’s former factory district transformed into a global hub for pop-ups, brand experiences and new forms of urban consumption. SEOUL, February 12 (AJP) - In Seongsu, a former industrial district in eastern Seoul, retail space is no longer leased mainly to sell goods. It is rented — by the day or the week — to buy attention. Over the past decade, short-term pop-up leases have replaced the two-year commercial contracts that still define most of the city’s retail landscape. In prime pockets such as Yeonmujang-gil, a large venue of about 900 square meters (around 270 pyeong) can now cost between 100 million and 200 million won ($75,000–$150,000) for a single week, according to 2024 market data from local brokerages and industry reports. If booked continuously, that translates into a monthly equivalent of 400 million to 800 million won ($300,000–$600,000). By contrast, the effective monthly rent (E.NOC) for traditional retail and office space in Seongsu stood at about 290,000 won per pyeong in 2023, according to the 2024 Seongsu Office Market Report by commercial real estate data firm Alsquare. The widening gap illustrates what analysts describe as a “decoupling” of asset value from retail fundamentals. A parallel market has emerged, where short-term “event leases” are priced not by expected shop sales, but by brand demand, seasonality and promotional timing. A Market That Runs on Weeks, Not Years This structure remains unusual in Korea, where stability has long been the defining feature of commercial leasing. “Contracts like those in Seongsu are very rare,” said Lee, a building owner who manages several properties in the capital. “Most places in Seoul are leased on a minimum two-year basis. Seongsu operates on completely different rules.” Those rules are shaped by flexibility — and by scarcity. “Most pop-ups here run on very short terms, from a single day to a few weeks,” said Kang, a Seongsu-based broker. “Everything is negotiable. But if you want to open for just one or two days, it’s extremely expensive. Daily rates often start from around 10 million won and rise quickly depending on demand.” For landlords, the appeal is clear. Short-term leases allow them to capture peak-season premiums that long-term tenants cannot match. For brands, the costs are justified as marketing investments rather than rent. A Global Outlier By international standards, Seongsu’s pricing has reached striking levels. In Tokyo’s Shibuya district, pop-up spaces typically start at around 920,000 yen ($5,700) per week for smaller venues. In New York’s Soho, prime Broadway retail rents average about $726 per square foot annually, translating to roughly $135,000 per week for a comparable 900-square-meter space. At its peak, Seongsu now rivals both. In many cases, transactions bypass conventional real estate brokers. Large venues operate as registered event spaces, signing short-term venue-hire agreements directly with brands, blurring the boundary between retail, exhibition and entertainment. For marketers, the rationale is straightforward. “Pop-ups are not about making money from sales,” said Lee Seung-hwan, head of FIG, a creative agency. “They are about branding. Once you add production, design and staffing, budgets can easily reach hundreds of millions of won.” The Economics of “Digital Bricks” Beyond weekly venue fees, interior construction alone often reaches about 1 billion won ($750,000) for high-end pop-ups — all written off over just two or three weeks of operation. Unlike conventional stores, where fit-out costs are depreciated over years, Seongsu pop-ups treat the entire investment as a one-off marketing expense. According to digital marketing firm Inquivix, the primary performance indicator is the “Offline-to-Online (O2O) Loop,” tracked through spikes in Naver searches, social media mentions and map “saves,” rather than in-store revenue. The physical space functions as a content studio — a place to generate what marketers call “digital bricks,” visual assets that build a brand’s online presence long after the pop-up closes. High Rewards, High Risk For smaller brands, however, the boom resembles a high-stakes gamble. “A lot of founders dream of opening a pop-up in Seongsu at least once,” said Yoon, a clothing brand owner in her 30s. “It’s a powerful way to introduce your brand. But everyone knows how expensive it is. Only companies with strong backing can really afford it now.” Rising costs have gradually narrowed participation to large domestic labels, global luxury brands and well-funded startups. According to the 2025 Pop-up Trend Report by Sweet Spot, the number of pop-up stores in its network jumped 109 percent in 2025 to 3,077, with Seongsu accounting for about 35 percent. What began as a fashion and beauty marketing tool has expanded into technology and finance. Data and crypto-related firms such as Palantir and Upbit have used Seongsu pop-ups to turn abstract services into physical experiences. Foreign Foot Traffic Fuels the Premium Foreign visitors have helped sustain the surge. Data from the Seongdong District Tourism Survey show that Seongsu’s foreign visitor count rose from about 60,000 in 2018 to roughly 3 million in 2024. Spending by foreign tourists in the district reached 74.8 billion won ($56 million) in 2024, up 1.8 times from a year earlier. “Even when daily rent runs into the millions of won, it can still make sense,” the broker said. “Foreign customers lift overall sales.” The leasing boom has fed directly into asset prices. Land values in Seongsu rose from about 40 million won per pyeong in 2018 to around 140 million won in 2023, according to Alsquare — more than a threefold increase. Analysts describe the result as a new phase of gentrification, in which long-term neighborhood businesses are pushed out by short-term marketing installations. For visitors, the spectacle often conceals the economics. “I just enjoy going to Seongsu to see what’s new,” said Lee Min-joo, a Seoul resident in her 30s. “I didn’t realize how much money goes into these places. But the interiors, scents and atmosphere stay in your memory. I almost map which brand was where.” That reaction captures Seongsu’s transformation. The district is no longer priced like a retail neighborhood. It is priced like media time — sold in short slots, bid up by competition, and justified by reach rather than receipts. 2026-02-12 16:26:34
  • LocknLock highlights new products, meets partners at Ambiente 2026
    LocknLock highlights new products, meets partners at Ambiente 2026 SEOUL, February 12 (AJP) - South Korean household goods maker LocknLock said on Wednesday it has completed its participation in Ambiente 2026, the world’s largest consumer goods trade fair, held in Frankfurt from Feb. 6 to 10, where it showcased new and flagship products to global buyers. The company said the event, which drew brands and buyers from 176 countries, served as a platform to present its portfolio across food storage containers, beverageware, cookware and child-friendly product lines. LocknLock added that its booth was designed to allow visitors to navigate each product category and that it hosted partner invitation programs and on-site events to engage buyers and visitors. At the exhibition, LocknLock introduced a range of products planned for 2026, including items from Daekket, a premium cookware brand it launched earlier. The company said Daekket focuses on products that link cooking and plating and has received positive market feedback since its debut. In the food storage segment, LocknLock showcased its Fresh Maestro Vacuum Container, which uses a three-stage vacuum sealing system, while in beverageware it presented the Metro Café Ceramic Tumbler, featuring an interior ceramic coating designed to preserve beverage taste. The company said the tumbler drew strong interest from visitors. LocknLock also displayed products already popular in the Korean market, including the Suit stainless frying pan and the Schoolfit Switch Color stainless steel bottle, which changes color depending on temperature. The company said it held meetings with around 40 partners during the exhibition to discuss operational plans for new products and global sales strategies for 2026, and also met potential new buyers to explore opportunities in new markets. LocknLock exports to about 120 countries through eight overseas subsidiaries, including in Vietnam, the United States and Germany, and has expanded its business beyond food storage containers into beverageware and small appliances, it said. “During the exhibition, we discussed operational plans for new products and global sales strategies for 2026 with key partners, and confirmed the potential to expand into new markets through meetings with new buyers,” said a LocknLock representative. 2026-02-12 15:32:51
  • Koreas Gentle Monster marks 15 years, grows into global eyewear brand
    Korea's Gentle Monster marks 15 years, grows into global eyewear brand SEOUL, February 10 (AJP) - South Korean eyewear brand Gentle Monster marked its 15th anniversary this month, underscoring its rise from a local label into a global fashion name as Korean brands gain traction overseas. The company currently operates 84 stores worldwide, including in New York, London, Paris and Milan, and has staged 33 collaborations with fashion houses and artists ranging from Fendi and Moncler to Maison Margiela and Tilda Swinton, according to the company. Gentle Monster, run by IICOMBINED, said overseas units now account for about 40 percent of its sales, highlighting growing demand for Korean-designed eyewear beyond its home market. Founded in 2011, the brand helped reshape Korea’s eyewear market by positioning glasses as a fashion item rather than a purely functional product, and has since expanded into major retail hubs across Asia, North America and Europe. “Over the past 15 years, we have grown into a global brand by focusing on product quality and a distinctive brand identity,” said a Gentle Monster spokesperson. “We will continue to expand into new areas beyond fashion through innovative design and differentiated customer experiences.” 2026-02-10 17:22:00
  • The next big chip from Korea – flash successor to HBM: KAIST expert
    The next big chip from Korea – flash successor to HBM: KAIST expert SEOUL, February 10 (AJP) - High-bandwidth memory (HBM) has powered the current AI boom, but soaring costs and capacity limits are shifting attention to what comes next. The next big chip to come from Korea's memory powerhouse will be flash-based successor designed for the inference-heavy phase of AI, according to a chip expert. “Now is the time to move beyond HBM. The era of high-bandwidth flash is coming — and it will unfold within the next decade,” Kim Jung-ho, a professor at Korea Advanced Institute of Science and Technology (KAIST), said Tuesday during a livestreamed session hosted by the KAIST Tera Lab, which he leads. Kim’s forecast reflects a growing industry concern that memory — not compute — is becoming the primary bottleneck in scaling next-generation AI systems. As large language models expand and inference workloads multiply, the challenge is no longer just faster processors, but how much data can be stored close enough to feed them efficiently. At the center of the problem is the explosive growth of so-called key-value (KV) caches — temporary memory that stores intermediate data during AI inference in transformer-based models. These caches expand rapidly with longer context windows, pushing HBM to its practical limits in both capacity and cost. HBM functions as ultra-fast “working memory” placed next to GPUs, but its capacity growth has lagged behind model demands. According to publicly available specifications from Nvidia, the H100 accelerator carries 80 gigabytes of HBM, while the H200 offers 141 gigabytes. Industry tracker TrendForce estimates the upcoming Blackwell-based B200 will reach roughly 192 gigabytes, with the next-generation Rubin platform expected to approach 288 gigabytes. Even those increases fall short. A single modern AI model — once intermediate data is included — can already consume hundreds of gigabytes of memory. Running a 70-billion-parameter model such as Llama-3 in FP16, for example, requires about 140 gigabytes just for model weights. When long-context inference is added, memory demand quickly exceeds what a single GPU can supply, limiting batch size and the number of users that can be served concurrently. The strain is more acute in transformer architectures, where KV caches scale linearly with context length. For a 70-billion-parameter model, KV cache requirements are estimated at roughly 45 to 64 gigabytes for a 128,000-token context window. At the million-token scale, that figure swells to roughly 327 to 512 gigabytes — two to three times larger than the model weights themselves. For 400-billion-parameter-class models, supporting million-token contexts would push memory needs into the terabyte range, far beyond today’s HBM-only configurations. “This is why performance is no longer determined by compute alone,” Kim said. “Once decoding begins, throughput becomes memory-bound. Capacity matters as much as bandwidth.” To bridge that gap, Kim argues that the industry will need high-bandwidth flash (HBF) — a stacked, NAND-based memory technology designed to offer far larger capacity than HBM while delivering much higher bandwidth than conventional storage. Rather than replacing HBM, HBF would serve as a complementary tier within a hierarchical memory system. Kim likens the relationship to a workspace: HBM is “the bookshelf next to your desk,” optimized for speed but limited in size, while HBF functions as “the library,” holding bulky data such as KV caches and long-context inputs that can be streamed back to the GPU when needed. The push toward longer context windows is already visible across the AI industry. Google’s Gemini 1.5 supports up to one to two million tokens, Anthropic’s Claude models target 200,000 to one million tokens, and Meta has signaled even longer contexts for future Llama models. Combined with retrieval-augmented generation and agent-based AI that accumulates session history, such workloads dramatically expand memory footprints. From a performance standpoint, HBM3e delivers roughly 1.2 terabytes per second of bandwidth per stack with latencies measured in tens of nanoseconds — but at a steep cost per gigabyte. Enterprise-grade solid-state drives, by contrast, offer only 10 to 30 gigabytes per second and latencies in the tens of microseconds, albeit at much lower cost. HBF aims to occupy the middle ground, targeting hundreds of gigabytes per second with microsecond-level latency, making it suitable for tiered AI memory architectures. Industry engineers increasingly expect future AI servers to adopt such layered designs, combining HBM, conventional DRAM and flash-based memory pools through high-speed interconnects such as CXL. In that setup, Kim said, HBM would remain dominant for latency-critical tasks, while HBF would emerge as the backbone for long-context and large-scale inference. “The question is no longer whether we need more memory,” Kim said. “It’s how we restructure the memory system itself. In the AI era, memory is becoming the real factory floor.” For memory makers, the shift opens a new front in the AI arms race — not only in faster chips, but in how much data those chips can store and move. 2026-02-10 16:29:07
  • Samsung Elec gets head start in HBM4 race, SK hynix plays it cool
    Samsung Elec gets head start in HBM4 race, SK hynix plays it cool SEOUL, February 09 (AJP) - Samsung Electronics has moved a step ahead in the race to commercialize next-generation high-bandwidth memory, shipping early samples of its HBM4 chips and preparing for mass supply, according to industry sources. In the sixth-generation memory standard expected to power future artificial intelligence accelerators from Nvidia and Google, Samsung appears to have gained an early edge over local rival SK hynix, which currently dominates the HBM market. A source familiar with the matter said Samsung completed sample shipments of HBM4 products to customers late last year and is now entering what amounts to the early commercialization phase. “Samples were already supplied by late December, and what comes next is essentially the process of shipping mass-produced products,” the source said. “Preparations for shipments are in the final stages.” The source added that delivery schedules remain closely tied to customers’ product road maps. “Even with detailed planning, schedules can shift depending on customers,” the person said. “But given the current situation, it would not be far off to view shipments as starting after the Lunar New Year holiday.” If Samsung begins commercial supply ahead of competitors, it would become the first memory maker to move HBM4 beyond the sample stage and into early customer shipments, industry officials said. Kim Deok Kee, a professor of electronic engineering at Sejong University, said Samsung would indeed be the first in the industry if it begins mass production ahead of its rivals. “There are only three major players in this market. Micron, Samsung Electronics, and SK hynix,” Kim said. “If Samsung moves first, it would be the first in the industry.” Kim said the early start carries symbolic weight for Samsung, which has lagged behind SK hynix in the HBM market. “Samsung has been playing catch-up in HBM, and taking the lead in HBM4 mass production could be seen as a chance to regain momentum,” he said. “It suggests the company has resolved key technical hurdles such as heat management.” SK Hynix Downplays Timing Gap SK hynix, however, has played down the significance of Samsung’s early move, saying it is already prepared for mass production and that supply timing ultimately depends on customers. “We announced last September that we had built a mass-production system after completing customer samples,” a company official said. “As for final supply, we and our competitors are at similar stages. It largely depends on customer timelines.” The company said it has finished sample deliveries and stands ready to move into full-scale supply once demand materializes, but declined to give a precise timetable. The contrasting messages reflect a shift in competition from technology development to commercialization speed, as chipmakers seek to align deliveries with Nvidia’s next-generation “Rubin” AI platform. Speed vs. Relationships Samsung is widely seen as leveraging its vertically integrated structure — spanning DRAM manufacturing, logic-die production and advanced packaging — to accelerate commercialization. HBM4 marks a major leap from the current HBM3E generation, doubling the input-output interface to 2,048 bits and sharply boosting bandwidth per stack. The upgrade is designed to ease the so-called “memory wall” that increasingly constrains AI system performance. SK hynix, meanwhile, continues to benefit from its long-standing partnership with Nvidia, built during the rapid expansion of HBM3 and HBM3E for AI workloads. Market Share Battle Intensifies SK hynix overtook Samsung as the world’s top memory supplier last year amid explosive demand for AI chips, driven largely by Nvidia’s platforms. In 2025, SK hynix generated 47.2 trillion won in chip operating profit on sales of 97.15 trillion won, compared with Samsung’s 24.9 trillion won. The three dominant memory makers — Samsung, SK hynix and Micron — which control most of the global memory supply, have sent memory prices across the board soaring as they increasingly shift capacity toward premium HBM products. Counterpoint Research estimates SK hynix will hold about 54 percent of the global HBM4 market this year, compared with 28 percent for Samsung and 18 percent for Micron. The firm previously put SK hynix’s HBM3 share at 62 percent in mid-2025. Investment banks also expect SK hynix to maintain its lead. Goldman Sachs forecasts the company will dominate HBM3 through at least 2026, while UBS estimates it could capture around 70 percent of HBM4 demand tied to next-generation AI platforms. Despite Samsung’s early push, SK hynix remains confident. “Our commercialization capabilities and product quality, which have earned customer trust, cannot be overtaken in a short period,” Song Hyun-jong, president of SK hynix’s corporate center, said during a recent earnings call. “As with HBM and HBM3, we aim to secure overwhelming leadership in HBM4.” Samsung Electronics shares closed at 166,400 won ($123), up 4.92 percent, while SK hynix ended at 887,000 won ($657), up 5.72 percent. 2026-02-09 15:42:56