So memory tax is effectively getting framed as Nvidia tax…
35%-40% of the total CAPEX will be spent on memory