NVDA Performance & AI Infrastructure Spend

It's becoming increasingly clear that a significant portion of NVDA's recent performance is tied directly to the massive capex expenditure on AI infrastructure. The question is how sustainable this growth rate is as hyperscalers build out their internal capabilities and other competitors in the custom ASIC space gain traction. Are we nearing peak 'build-out' for the immediate future? What are your thoughts on the next 12-18 months for core GPU demand outside of these mega-deals?