NVIDIA's upcoming "Rosa" CPU, initially expected on TSMC's 2nm family, could directly utilize the A16 (1.6nm) process to leverage Super Power Rail backside power delivery technology for higher transistor density. This shift to A16 would significantly increase demand for Chemical Mechanical Planarization (CMP) processes and consumables, requiring an additional 15%-30% more polishing steps than the N2 node.
TSMC's A16 process, if adopted by NVIDIA, will tighten supply for advanced CMP consumables and equipment, increasing manufacturing costs for leading-edge chips.