South Korea-based semiconductor giant SK hynix has announced the start of mass production of its next-generation HBM4 memory, specifically developed for AI applications. The company states that this new memory will revolutionize high performance and energy efficiency.

New HBM4 memory begins mass production

SK hynix’s HBM4 memory was designed to address the rapid growth in AI demand and the increasing data processing needs. The new memory doubles the bandwidth by doubling the number of input/output (I/O) terminals compared to the previous generation. It also provides a more than 40 percent increase in power efficiency.

According to SK hynix’s own calculations, the use of HBM4 can increase AI service performance by up to 69 percent and significantly reduce data center energy costs.

The company minimized risks in mass production by using its market-proven Advanced MR-MUF process and fifth-generation 10-nanometer (1bnm) technology in HBM4 production. HBM4 memory exceeds the JEDEC standard operating speed of 8 Gbps, reaching data throughput speeds exceeding 10 Gbps.

Justin Kim, president of AI infrastructure at SK hynix, stated that the start of mass production of HBM4 is a symbolic milestone that transcends the boundaries of AI infrastructure. Kim added that HBM4 will be a key product in overcoming technological challenges and will rapidly transform the company into a full-service AI memory provider by providing high-quality and diverse memory products.

So, what are your thoughts on this? Share your thoughts with us in the comments section below.