Semiconductor News 20260402

Today’s semiconductor news is a vivid mix of massive investment in next-generation technologies to meet the surging demand for AI infrastructure and geopolitical moves to strengthen the supply chain. Dramatic progress is being made in building the manufacturing infrastructure for cutting-edge processes in Japan, including plans by Fujitsu and Rapidus to produce 1.4nm AI chips in Japan and approval for 3nm volume production at TSMC’s Kumamoto Plant 2. Also, technological innovations driving the AI ecosystem are accelerating, such as NVIDIA’s $2 billion investment in Marvell and Micron’s development of stacked GDDR. At the same time, strategic defensive measures in response to material supply concerns and the reorganization of global supply networks, such as the expansion of helium stockpiling by Korean manufacturers and Intel’s acquisition of a wholly owned Irish factory, are also prominent, and market growth and risk countermeasures are occurring simultaneously.

Main News

1. Fujitsu Plans Domestic Production of 1.4nm AI Ch ips in Collaboration with Rapidus
   Fujitsu has announced plans to collaborate with Rapidus, a cutting-edge Japanese semiconductor manufacturer, to produce AI chips in Japan using the next-generation 1.4nm process. This is a strategic move by a major Japanese IT company and a foundry to team up directly at the forefront of miniaturization technology, aiming to break away from overseas dependence and strengthen the domestic ecosystem. This is an important step toward the domestic production of cutting-edge AI processors.
2. Taiwanese Authorities Approve 3nm Process Manufacturing at TSMC’s Second Kumamoto Plant, Mass Production to Begin in 2028
   Taiwanese authorities have officially approved TSMC’s plan to manufacture semiconductors using the cutting-edge 3nm process at its second plant to be built in Kumamoto Prefecture, Japan, with mass production scheduled to begin in 2028. This will ensure that TSMC will have a manufacturing base in Japan for cutting-edge processes essential for AI semiconductors and other products, and is expected to have a significant ripple effect on the resilience of the Japanese semiconductor supply chain and related industries.
3. NVIDIA to invest $2 billion in Marvell to cooperate in silicon photonics and other areas
   NVIDIA, a leader in the AI semiconductor market, will invest a massive $2 billion in Marvell, a U.S. company that will link GPUs to custom ASICs and eliminate data transfer bottlenecks. strong collaboration in areas such as the use of silicon photonics technology. With this major investment, Marvell aims to form a huge ecosystem involving its competitors and surround them with next-generation technologies that will fundamentally raise the performance of AI infrastructure.
4. South Korean Semiconductor Makers Stockpile Helium to Diversify Supply Network
   Major South Korean semiconductor manufacturers are rapidly expanding their stockpiles of helium, a special gas essential for semiconductor manufacturing. Against the backdrop of unrest in the Middle East and other factors, moves to reduce dependence on Qatar and diversify supply routes from other producing countries such as the U.S. and Australia are gaining momentum. Against the risk of physical supply disruptions of key materials, defensive procurement strategies to protect the stable operation of manufacturing are being implemented at a rapid pace.
5. Micron is Developing Stacked GDDR to Meet Changing AI Demand
   Memory giant Micron is reportedly developing stacked GDDR to meet diversifying memory demand for AI. This is a move to pioneer a new cost-effective architecture suitable for edge AI and inference processing as an alternative or complement to expensive and tight supply of high-bandwidth memory. Memory products are becoming increasingly fragmented to meet the various needs of the rapidly expanding AI chip market.
6. Intel to Expand Advanced Semiconductors by Making Irish Joint Venture Factory Wholly Owned
   Intel has announced that it will buy back the remaining 49% equity interest in its semiconductor manufacturing fab joint venture in Ireland, making it a wholly owned subsidiary. The move is intended to independently expand production capacity and shipments of advanced semiconductors in the European market and to reorganize and strengthen control of its global manufacturing footprint. This is a strategic investment to diversify the supply chain against a backdrop of geopolitical risk and to ensure competitiveness in Europe.
7. South Korea’ s March Semiconductor Exports Surge 151.4% YoY to Top $30 Billion
   South Korea’s semiconductor exports in March surged an astounding 151.4% YoY, breaking through the $30 billion barrier in a single month for the first time in history, according to a report released today. The explosive growth in demand is believed to be the main factor. The memory market, which had been stagnant for a long time, is now on a full recovery track, and the strong demand is driving the export front of the Korean economy strongly.
8. Epson and Rapidus Collaborate to Apply Printing Technology to Semiconductor Manufacturing
   It has been reported that Seiko Epson and Rapidus will collaborate in the semiconductor manufacturing process. This is an attempt to apply the inkjet and other high-precision printing technologies that Epson has developed over many years as a new approach to semiconductor manufacturing. This is an innovative approach that aims to significantly reduce manufacturing costs and environmental impact by replacing some of the complex manufacturing processes of the past, and is attracting attention as a collaboration that leverages the strengths of the Japanese manufacturing industry.
9. DDR5 and Other Memory Prices Fall in Chinese Market, Weakening Supply/Demand
   In the Chinese market, we reported that DDR5 prices are falling, driven by module inventories in the secondary market. In addition, overall spot prices for memory are weakening due to weak demand for DDR4 and DDR5, while DDR3 is firm in the high end of the price range, and while high-bandwidth memory for AI is booming, local supply-demand slack and price adjustments in the general-purpose memory market are surfacing.
10. NVIDIA Aims to Dominate Ent ire AI Stack with Three System Deployments
    NVIDIA has revealed its strategy to go beyond simply providing GPU hardware to dominate the entire AI stack through three system deployments. This is a bold move to comprehensively dominate the computing infrastructure of the AI era.

News List