后摩智能确认其下代大模型端边 AI 芯片采用 3D CIM 存算一体架构

AI Digest

后摩智能公布下代大模型端边AI芯片采用3D CIM存算一体架构，通过垂直堆叠提升算力并优化功耗，展现AI芯片技术突破。

Houmo.AI announces next-gen edge AI chip using 3D CIM architecture for enhanced compute efficiency, marking AI chip innovation.

Key points

• 下代芯片采用3D CIM存算一体架构提升算力 Next-gen chip uses 3D CIM architecture for compute boost
• 垂直堆叠技术实现同等面积下2倍算力 Vertical stacking achieves 2x performance in same area
• 融合3D DRAM技术优化功耗与散热 3D DRAM integration optimizes power/thermal management
• M50芯片已实现160TOPS算力量产应用 M50 chip delivers 160TOPS production capability
• 覆盖超150款产品定点的多场景部署 Over 150 product deployments across scenarios

Takeaway: 3D CIM架构突破将重塑端边AI芯片性能边界。 / 3D CIM architecture breakthrough redefines edge AI chip performance.

Why it matters 展示AI芯片架构创新对算力效率的革新影响，具行业前瞻价值。