Stealth Startup Kepler Computing Unveils New Memory Architecture to Ease Chip Shortage

Kepler Computing, a semiconductor startup founded by physicists and computer scientists in 2018, announced its exit from stealth mode with a new memory architecture that combines 3D stacking and a low-voltage ferroelectric composite to increase SRAM and high-bandwidth memory density without EUV lithography. Backed by $468 million from GlobalFoundries, Intel Capital, AMD Ventures, Baillie Gifford and Bill Gates’ Gates Frontier fund, the firm also received a commitment of up to $245 million from the U.S. Department of Commerce for U.S. development.

Prototype production is occurring at GlobalFoundries facilities in Singapore and Burlington, Vermont, where the startup has already processed about 2,000 wafers. Kepler aims to ship its first HBM samples this year, scale Singapore output next year, and begin U.S. manufacturing by 2028, betting that retrofitting existing fabs is cheaper than building new multibillion-dollar plants. Executives stress that the approach could cut energy use by bringing compute closer to memory while preserving HBM’s capacity. Industry observers note that scaling such material-based innovations remains a major hurdle.