Samsung and SK hynix Boost First-Half Semiconductor Spending by 35% as NVIDIA Misses Top Customer List

Deep News08-17 20:03

Driven by surging demand from the artificial intelligence sector, Samsung Electronics Co., Ltd. and SK hynix have aggressively expanded their production capacity, with combined semiconductor facility investments in the first half of 2026 climbing by more than 30% year-on-year.

According to the companies' semi-annual reports for 2026, as cited by South Korean media outlet Etoday on August 17, the two tech giants' total semiconductor facility spending reached 43.198 trillion Korean won during the January-to-June period, a 35.1% increase from 31.98 trillion won in the same period last year. Both firms reported 100% capacity utilization, underscoring the exceptional strength of current AI memory demand.

Regarding customer composition, revenue contributions from NVIDIA to SK hynix fell to 13.35% from 24% for the full year of 2025, while NVIDIA did not even appear on Samsung Electronics Co., Ltd.'s list of top five clients during the same timeframe. This divergence highlights the differing strategic approaches the two companies are taking in their AI business operations.

Robust AI Demand Fuels Accelerated Expansion by Memory Leaders

In terms of investment pace, SK hynix has adopted a more aggressive expansion strategy. Samsung Electronics Co., Ltd.'s DS division increased its facility investment by 23.5% year-on-year in the first half, directing funds primarily toward next-generation memory, advanced process capacity, and related infrastructure. Meanwhile, SK hynix saw a substantial 56.4% jump in facility spending, focusing on meeting demand for HBM, high-capacity DRAM, and enterprise-grade SSDs.

Both companies are operating at full capacity, indicating that existing production lines are essentially running at maximum output. With ongoing AI infrastructure development driving rapid growth in demand for premium memory products like HBM, Samsung Electronics Co., Ltd. and SK hynix are proactively expanding capacity for next-generation offerings ahead of time.

In other words, this round of capital expenditure is not merely about keeping pace with current demand but is aimed at securing a competitive edge in both production capacity and technology for the next wave of AI memory growth.

Record R&D Investment as SK hynix Nearly Doubles Spending

Beyond capital expenditure, both companies are simultaneously ramping up their research and development efforts. Samsung Electronics Co., Ltd. recorded total company-wide R&D spending of 27.363 trillion won in the first half, up 51.5% year-on-year and marking a record high for any half-year period. However, the company did not disclose separate R&D figures for its DS division. SK hynix showed even more striking growth in R&D investment, reaching 6.043 trillion won in the first half, a 98.4% surge year-on-year, which is equivalent to roughly 90% of its total R&D spending for all of 2025.

The competition among memory manufacturers is also extending beyond traditional battles over process technology and capacity into the broader AI computing ecosystem. South Korean media reports indicate that Micron recently launched the Micron Ventures Paradigm Fund with a size of $250 million, the company's largest venture capital fund to date, targeting areas such as AI model architectures, computing, enterprise applications, and physical AI. The move is designed to position the company early for emerging AI technology trends and to build a network of partnerships related to next-generation memory and computing products.

AI Memory Demand Shifts from NVIDIA Reliance to Diversification

According to the semi-annual reports, changes in customer structure stand out as one of the most notable developments of the period. As reported by the Seoul Economic Daily, SK hynix's revenue from NVIDIA in the first half of 2026 reached 17.6087 trillion won, accounting for 13.35% of total revenue, down from the 24% share recorded for the full year of 2025.

This shift does not necessarily signal a decline in NVIDIA's importance. On the contrary, SK hynix is expanding its supply to major tech companies developing their own AI chips, resulting in a more diversified customer base. As SK hynix progresses with HBM4 supply in conjunction with NVIDIA's Vera Rubin platform, the contribution from NVIDIA to its revenue is expected to recover in the second half of the year.

Samsung Electronics Co., Ltd.'s customer structure, by contrast, shows a higher degree of concentration. According to its semi-annual report, the top five clients together accounted for approximately 25% of total revenue, with NVIDIA failing to make the list.

This suggests that as demand for AI chips surges, the relationships between memory manufacturers and their clients are evolving: NVIDIA remains one of the most critical customers in the HBM market, but with tech giants like Alphabet and Amazon accelerating their in-house ASIC development, the sources of AI memory demand are becoming increasingly diversified.

For Samsung Electronics Co., Ltd. and SK hynix, the competition ahead is no longer just about who can produce more HBM, but also about who can quickly penetrate the expanding pool of AI chip customers and secure more core clients for next-generation products like HBM4.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.

Comments

We need your insight to fill this gap
Leave a comment