NVIDIA should remain the default
AI factory stack through 2031, allowing revenue to more than double via
Rubin-era compute, networking and software attach; the upside increasingly comes from owning deployment speed, trust and workflow control points, not just selling scarce chips.