SK hynix Leads the Charge in Next-Gen AI Memory with World's First 12-Layer HBM4 Samples

SK hynix Unveils Groundbreaking 12-Layer HBM4 Memory

In a significant leap forward for AI memory technology, SK hynix has announced the successful sampling of the world's first 12-layer High Bandwidth Memory 4 (HBM4) to major customers 1

1

. This development marks a crucial milestone in the ongoing race to enhance AI acceleration capabilities, with SK hynix solidifying its position as a leader in the AI memory market.

Unprecedented Performance and Capacity

The new 12-layer HBM4 samples boast impressive specifications that set them apart in the industry:

• Bandwidth: Capable of processing over 2 terabytes (TB) of data per second, equivalent to processing more than 400 full-HD movies in just one second 2

  2

  .
• Speed Improvement: More than 60% faster than the previous generation HBM3E 3

  3

  .
• Capacity: Achieves an industry-leading 36GB capacity, the highest among 12-layer HBM products 4

  4

  .

Advanced Manufacturing Process

SK hynix has implemented its Advanced MR-MUF (Mass Reflow-Molded Underfill) process in the production of HBM4. This innovative technique:

• Prevents chip warpage
• Maximizes product stability
• Improves heat dissipation 2

  2

Accelerated Timeline and Market Impact

The company has expedited its HBM4 development and production schedule:

• Samples delivered ahead of the initial plan
• Mass production preparations targeted for the second half of 2025
• Original mass production was set for 2026, but timeline moved up by approximately six months 1

  1

Industry Collaboration and Future Prospects

SK hynix's HBM4 development is closely tied to the needs of major players in the AI industry:

• NVIDIA is expected to use HBM4 in its upcoming Rubin series of GPUs 3

  3

  .
• The company aims to strengthen its position in the next-generation AI memory market 2

  2

  .

Broader Industry Trends

The development of HBM4 is part of a larger trend in the memory industry:

• Other major players like Samsung and Micron are also working on HBM4 solutions 5

  5

  .
• Future iterations like HBM4e are already being planned, with potential for even higher capacities and speeds 5

  5

  .

As AI workloads continue to demand more from memory systems, the race to develop faster and higher-capacity solutions like HBM4 is likely to intensify, driving innovation in the semiconductor industry and enabling more powerful AI applications in the coming years.