On August 19, SemiMedia reported a significant increase in DDR4 prices, which has boosted the performance of memory module suppliers and DRAM manufacturers. Tight supply and delayed technological transitions have extended the market momentum through the end of the year. Adata’s earnings per share reached NT$2.75 in the second quarter, with a 39.5% year-on-year increase in revenue in July. Etron Technology saw a 22.4% month-on-month increase in revenue in July. Nanya raised contract prices by 11% to 16% in August, with a 31.4% month-on-month and 94% year-on-year increase in revenue in July.
On August 22, according to TechNews, citing the Nikkei, Japan has long been a leader in the power semiconductor field, but the rapid rise of Chinese companies is posing a serious challenge. Japan’s slow progress in industry consolidation and strategic coordination has limited its ability to compete with Chinese companies.
The report pointed out that the technological gap between Japanese and Chinese companies in silicon chips is only one to two years, and at most three years in silicon carbide (SiC) chips. Japan has five major power semiconductor manufacturers, but each has a global market share of less than 5%. Cooperation and integration among these companies often face obstacles.
For example, in 2023, Rohm invested 300 billion yen in Toshiba to complement resources and expand into electric vehicle and industrial applications, but broader cooperation has not yet been realized. Rohm reported a net loss of 50 billion yen in the fiscal year ending in March 2025, its first annual loss in 12 years. The company struggled to profit from next-generation SiC power chips amid a slowdown in the electric vehicle market, while fierce competition from Chinese rivals further affected its earnings.
On August 21, SemiMedia reported that Samsung’s global DRAM market share dropped to 32.7% in the first half of 2025, a significant decline from 41.5% in the same period last year, falling below 40% for the first time and raising concerns about its leadership in the memory field.
Samsung’s DRAM dominance, which peaked at 48% in 2016, has been steadily declining. The recent decline is mainly due to weak demand for HBM, which is crucial for artificial intelligence processors. Samsung is currently undergoing qualification tests for HBM orders with major customers.
Samsung invested 18 trillion won in R&D in the first half of the year and filed nearly 10,000 patents, highlighting its determination to strengthen its semiconductor competitiveness through innovation. The decline in DRAM share highlights the pressure on its memory business, with the uncertainty of HBM adoption being the biggest challenge. Diversification and continued R&D are key to defending its position in the global semiconductor industry.
On August 21, the Times reported that Adata held an online earnings call on Thursday, expressing optimism about the DRAM market from the second half of this year to next year, expecting a strong return of DDR4 demand in September. Adata’s third-quarter performance is expected to be on par with the second quarter, with a relatively optimistic outlook for the second half of the year and better profits than the first half.
Adata stated that DRAM is in short supply due to the rapid growth in AI demand, with a positive market trend; although NAND supply dynamics are more complex, supplier discipline has also led to a relatively strong market. Despite demand adjustments for DDR4 in July and August, it is expected to return strongly in September, with third-quarter performance expected to be on par with the second quarter. Adata’s inventory level at the end of the second quarter was NT$1.44 billion, maintained at a 3-4 month level, with adjustments based on actual conditions.
Regarding rumors that Samsung has delayed the shutdown of EOL, Adata stated that Samsung has not delayed EOL, and customer DDR4 applications and demand remain as planned. From the fourth quarter to the first quarter of next year, DDR4 demand and prices are expected to show a positive trend and continue to rise. Next year, DDR5 supply and demand are still tight, with an overall positive DRAM market trend. If NAND market conditions cooperate, Adata’s operations next year are expected to maintain a positive development.
According to a report by the United Daily News, on August 19, Jason Chiang, General Manager of Foxconn’s Business Group B and Digital Health, said that Foxconn plans to produce AR glasses targeting the consumer market, with mass production as early as next year, aiming to become a leader in the next-generation human-computer interaction platform.
Foxconn announced a cooperation with Nobel Optometry Technology in three areas: the AI Agent-empowered fundus image analysis system, near-eye display technology (NED), and the cross-domain Taiwan Digital Health Industry Alliance, to promote the intelligence and internationalization of Taiwan’s visual medical field. Chiang pointed out that the key to the success of AR glasses is visual comfort. Foxconn needs to cooperate with the medical system to obtain professional eye information and provide a comfortable and safe wearing experience.
Chiang emphasized that Foxconn’s strategy in the AR glasses field is not limited to contract manufacturing but aims to build a complete ecosystem from components, modules, to systems, becoming a market leader.
On August 20, according to industry media reports, as competition in the global semiconductor foundry industry intensifies and chip manufacturers rush to advance 2nm technology, TSMC has priced its 2nm wafers at around $30,000 per wafer, implementing a “no discount, no negotiation” policy, a 50% to 66% increase compared to its current 3nm products.
TSMC plans to begin initial production within the next three years, with an initial monthly capacity of 30,000 to 35,000 wafers, expected to expand to around 60,000 wafers per month by 2026 with the launch of new fabs. The initial yield rate is expected to be around 60%, while the SRAM yield rate is expected to exceed 90%, sufficient to support mass production. Analysts pointed out that this pricing strategy aims to reflect costs and allocate the limited initial capacity to high-value markets such as high-performance computing (HPC) and artificial intelligence (AI).
