Preferred Networks Develops MN-Core L1000 AI Semiconductor Optimized for Deep Learning

Preferred Networks, Japan's leading AI unicorn, has begun developing MN-Core L1000, an AI processor optimized for generative AI and large language models. Announced in November 2024 with delivery targeted for 2026, the L1000 achieves up to 10x faster performance than existing GPUs during inference. The deep learning-specialized processor uses 3D-stacked DRAM technology to enhance memory bandwidth and performance efficiency.