Weebit Nano takes aim at costly AI challenges of more speed and less power | Stockhead

SK hynix says the next major bottleneck for AI won’t just be making computers faster, but reducing the power they consume, warning at a Seoul technology conference that the constraint is shifting from computing performance to memory and electricity. The company is working on next-generation memory aimed at cutting energy wasted when moving information within AI systems. The story links that challenge to the growing role of inference: McKinsey estimates that by 2030, around 60% of AI computing demand could come from inference rather than training, meaning efficiency gains could compound. In that context, ASX-listed Weebit Nano (ASX:WBT) is developing resistive RAM (ReRAM) and exploring in-memory computing—performing certain calculations inside memory—to improve speed and energy use.