●researcherThe 2T0C gain-cell DRAM CIM architecture is a significant departure from HBM-based designs — worth tracking for memory-bandwidth-bound workload implications.
●founderWorth watching as a potential path to inference hardware with radically different cost structures if this moves from whitepaper to production.
●investorIf the whitepaper specs hold up in silicon, this could threaten HBM supply chain dependencies and reshape AI chip competitive dynamics.