So far, we've seen Micron and Samsung list their next-generation HBM4 memory products while confirming development. Both companies have emphasized a time frame of launching products around 2025-2026, and it seems SKhynix has also joined the battle for time and technology.
As the adoption of AI rapidly increases in the market and we require greater computing power as we move into the future, it is important to note that HBM plays a vital role in how AI computing positions itself in modern life, as it is a key component in manufacturing AI accelerators.
During a keynote speech at SEMICONKorea 2024, SKhynix Vice President Kim Chun-hwan revealed that the company intends to start mass production of HBM4 by 2026, claiming that this will drive huge growth in the artificial intelligence market.
He believes that in addition to enabling an early transition to next-generation products, it is important to recognize that the HBM industry is facing a huge demand; therefore, it is even more important to create a solution that is both seamlessly supplied and innovative. Kim believes that the HBM market is expected to grow by up to 40% by 2025 and has been positioned in advance to take full advantage of this market.
Regarding expectations for HBM4, the roadmap shared by Trendforce predicts that the first HBM4 samples will have up to 36GB per stack capacity, while the full specifications are expected to be released by JEDEC around the second half of 2024-2025. First customer samples and availability are expected in 2026, so it’s still a long time before we see the new high-bandwidth memory AI solution in action.
It is still uncertain which types of AI products will use the new process, so we cannot make any predictions at the moment. With the addition of SKhynix, the competition in the HBM market seems to become more intense. Let us wait and see which company will rise and ascend to the throne of the king.