SoftBank and Intel Team Up to Engineer Next-Gen AI Memory Tech

SoftBank plans to invest up to 3 billion yen (approximately $19 million) over the next two years to kickstart the project, according to sources familiar with the matter. While total research and development costs are still being finalized, the primary objective is to create memory systems capable of supporting the massive data throughput required for training large-scale AI models while significantly reducing the electrical load on infrastructure.

Disrupting the Memory Status Quo

The venture enters a market currently dominated by a trio of giants: Samsung Electronics, SK Hynix, and Micron. Analysts suggest that for a newcomer to effectively compete, strategic alignments with established AI powerhouses will be essential. Jukan Choe, an analyst at Citrini Research, noted that the SoftBank-Intel partnership would likely need to collaborate with a market leader like Nvidia to truly shift the industry landscape.

The initiative also aligns with broader economic security goals in Japan and the United States. As governments scramble to secure resilient semiconductor supply chains, the partnership seeks to alleviate the supply squeeze caused by the explosive growth of AI firms. By focusing on next-generation manufacturing, the companies aim to provide a stable alternative to the existing high-bandwidth memory bottleneck.