MILLIWATT TO MEGAWATT

Building 3D Blocks→Different Technology Stacks

Each Whalechip technology is a different composition of the same Veltra-W building blocks. As workload scale grows, more 3DIC blocks are added — memory tiers stack higher, compute integrates deeper.

WoW DRAMNPU+RISC-VLogicWoW SRAMGPUHeterogeneousSwitchActiveInterposerBELUGAORCALEVIATHANIn DevelopmentPlanned

3D Edge AI · Milliwatt

BELUGA

Ultra-low-power WoW stacked compute for edge inference and vision processing. Built on WoW CUBE DRAM + NPU/RISC-V Logic, optimised for constrained environments.

3D Mid-Range · Multi-Watt

ORCA

Balanced compute-memory integration with on-chip SRAM cache and WhaleSwitch networking fabric for industrial AI and mid-range HPC workloads.

3D Data Center · Megawatt-Class

LEVIATHAN

Extreme-bandwidth multi-stack architecture deploying all six building blocks — including GPU heterogeneous compute and Active Interposer — for AI training and cloud-scale HPC.