South Korea’s SK Hynix, the world’s second-largest memory chip manufacturer, has announced it will begin mass production of its next-generation HBM4 (High Bandwidth Memory) chips in 2026, marking a pivotal step in the evolution of high-performance computing and AI acceleration.

🧠 HBM4 Targets AI and Data Centers

SK Hynix confirmed it is working closely with major global clients—including Nvidia, Intel, and AMD—to develop the new memory solution tailored for high-speed computing environments such as generative AI models, LLMs, and data center GPUs.

The HBM4 technology is expected to double the bandwidth of the current HBM3E, reduce power consumption, and offer unprecedented integration with cutting-edge processors.

🔧 Technology Under Development

The company revealed that engineering samples of HBM4 will be shipped to partners in late 2025, with full-scale mass production beginning in early 2026.

SK Hynix is currently finalizing its 12-stack and 16-stack chip architectures, which will increase bandwidth to over 1.5 TB/s. Additionally, it plans to improve thermals using advanced thermal interface materials and through-silicon via (TSV) technology.

🌐 Industry Race Heats Up

The HBM market has become a hotbed of innovation, driven by skyrocketing demand for AI training chips. Rivals Samsung and Micron are also racing to commercialize HBM4 by 2026. However, SK Hynix holds the advantage of having dominated over 50% of the HBM market share as of 2024.

According to TrendForce, the global HBM market is expected to grow by 45% annually, largely fueled by next-generation AI models, smart factories, and autonomous systems.

🏭 Strategic Expansion & AI Momentum

To support the upcoming HBM4 mass production, SK Hynix is investing in expanding its M15X and M16 fabs in Cheongju and Icheon, respectively. The company has also hinted at further capacity expansion in North America or Southeast Asia.

CEO Kwak Noh-Jung stated, “HBM4 is not just an upgrade—it is our commitment to shaping the next era of computing. We’re enabling AI companies to push the boundaries of what’s possible.”