As the demand for AI chips continues to rise, the storage and server supply chains are also accelerating their adjustments. Reports indicate that SK Hynix is in contact with Intel, considering producing storage chips in the United States; on the other hand, Apple is also said to be evaluating the launch of AI servers for external customers.
SK Hynix considers expanding its storage production capacity in the US
Reuters cites sources stating that SK Hynix is in talks with Intel, possibly to rent one of Intel's factories located in Ohio for the production of memory chips in the United States. SK Hynix is currently building a chip packaging factory in Indiana, but has not previously produced memory chips on U.S. soil.
SK Hynix is the world's second-largest manufacturer of memory chips, ranking alongside Samsung Electronics and Micron as major memory manufacturers. The company is also an important supplier to American technology companies such as Apple, Google, Meta, Microsoft, and NVIDIA.
HBM's whereabouts remain the biggest highlight
If negotiations progress, this will add new production capacity to the supply of semiconductors in the United States and also help alleviate the current shortage of memory chips. For Intel, which is in a period of business adjustment, leasing out factory capacity could also lead to additional utilization.
It is currently unclear which type of products SK Hynix plans to produce in the United States. The company's products cover DRAM and NAND flash memory, but what the market is most concerned about is still HBM, which is high-bandwidth storage. Such products are widely used in AI chips and are also one of the businesses with the most prominent profits in recent years.
However, this arrangement may also come under scrutiny from South Korea. Reports suggest that if South Korean core semiconductor companies move their key production capacity overseas, Seoul may not be pleased with that, especially against the backdrop of ongoing changes in the US semiconductor tariffs policy.
Apple is reported to be planning to develop its own AI servers
The Information reports that Apple is developing AI servers using its own custom chips and is considering integrating NVIDIA's NVLink interconnection technology, aiming to sell them to AI developers, enterprises, and government customers.
The report mentions that Apple may be preparing two versions: one with two M8 Ultra chips and another with four M8 Ultra chips. The tentative timeline for these products points to 2029.
Apple is currently building enterprise-level servers internally to support its global AI services. If it further commercializes such capabilities, it means that its role will extend from being a provider of terminal devices and cloud services to the AI infrastructure market, and it may face more direct competition with Supermicro, Dell, HPE, and NVIDIA.
Additional information:The first half of the original text, which discusses American voters' survey regarding AI and data centers, as well as the second half, which covers EU child online legislation and other tech updates, are not the main focus of this article and have been removed in accordance with journalistic guidelines.
