SK Hynix unveiled its iHBM technology on May 26, embedding an integrated cooling element in HBM packages. As AI computing demand rises and the number of HBM stacks and speeds increase, heat control has emerged as a next-generation memory technology issue. The company adopted an approach that secures a heat-release path within the package structure itself.

Conventional HBM has relied on an indirect method that sends heat out through the core die. The area where heat is most concentrated is the D2D PHY (Die-to-Die Physical Layer), a physical connection between the HBM base die and the AI accelerator die. iHBM inserts ICE (Integrated Cooling Elements) as a heat-control component in this section to create a separate, dedicated path for heat to escape. ICE is made of silicon material that is not electrically conductive but has high thermal conductivity.

The company said the structural change can reduce thermal resistance by more than 30 percent compared with existing designs. It said the technology can maintain stable operating characteristics even in high-temperature, high-load environments. Heat density refers to the amount of heat generated per unit area and is cited as an indicator that determines cooling efficiency and system lifespan.

On mass production, it applied a market-proven wafer-level packaging process based on advanced MR-MUF. MR-MUF injects a protective material into the spaces between stacked chips and hardens it. The company said the technology can be applied without major additional design changes by securing high design compatibility with customers' existing SiP environments.

SK Hynix plans to apply iHBM sequentially starting with next-generation products such as HBM5. It aims to meet heat-management standards required in ultra-high-density and ultra-high-bandwidth environments such as high-performance computing and AI data centres, and to enhance overall system stability and operating efficiency.

Lee Kang-wook (이강욱), a vice president at SK Hynix in charge of PKG development, said, "iHBM is the optimal solution to minimise heat generation, developed by combining memory design capabilities and advanced packaging technology." He added, "We will proactively provide the value customers need in the AI environment and further solidify AI memory leadership."