Having observed the Hopper & Blackwell and now the Rubin ramp over the last 4 years, plz don't expect Nvidia to meet their deadlines
Jensen operates on very aggressive timelines and puts too much pressure on the supply chain
AI servers are much more complex than smartphones (the logic of the 2010s won't work here) and are also undergoing rapid iteration every generation (HBM, CPO, AP, 800 VDC, substrates/PCB size and complexity etc)