A California-based startup, Kepler Computing, has emerged from stealth mode after seven years, claiming its innovative memory chip architecture could alleviate the global memory chip shortage. The company's approach bypasses the need for expensive extreme ultraviolet lithography (EUV) by utilizing 3D stacking and a proprietary new material.
Founded in 2018 by physicists and computer scientists, Kepler Computing has developed a new architecture for high-bandwidth memory (HBM) and high-speed cache memory (SRAM). Their method aims to increase chip density without relying on EUV, making it compatible with existing semiconductor fabrication plants. This is a significant development as chipmakers typically use EUV to shrink transistors, packing more technology into smaller spaces.
Kepler has secured substantial funding, totaling $468 million from prominent investors including GlobalFoundries, Intel Capital, AMD Ventures, Baillie Gifford, and Bill Gates. The US Department of Commerce has also committed up to $245 million to support Kepler's development of advanced AI memory technology in the United States.
The company's strategy involves a two-pronged approach. For HBM, Kepler employs a novel 3D-manufacturing technique to enhance density and reduce data transfer energy consumption. For SRAM, they have improved density using ferroelectrics and a new low-voltage composite material, which allows for data reading and writing at lower voltages.
While Kepler has made significant strides, the path to full-scale production involves challenges. The company plans to ship its first HBM chip samples later this year, ramp up production in Singapore next year, and begin chip production in the US by 2028. A key hurdle is managing the integration of materials like iron, which can be a contaminant in production facilities, requiring dedicated equipment or robust encapsulation.
Kepler Computing's ambition highlights a broader trend of startups seeking to innovate within semiconductor manufacturing to overcome bottlenecks. The success of their approach will hinge on their ability to scale production efficiently and cost-effectively, potentially offering a much-needed solution to the ongoing demand for advanced memory chips.
