Hyperscaler consensus assumes revenue growth stays at the same level as the last few years, but capex stops going up.
A few thoughts:
If GPus are what’s driving rev growth, then mix shift within capex spend towards gpu away from shell/labor/power (seems unlikely) could drive growth absent capex stepups. I don’t think there are a ton of empty shells out there tho, so that seems unlikely.
If Revenue growth wasn’t driven by GPUs and will accelerate for some other reason, then consensus could be right. Unclear what exactly that is, so seems unlikely.
If revenue growth was driven by GPUs (ie, META/GOOG have said they see returns to GPU spend = compute capacity drives rev growth) , then to get future rev growth you need GPU spend and then consensus is wrong, and either hyperscaler rev growth continues and GpU spend matches OR hyperscaler rev growth slows and GPU spend falls,
Don’t think ASICs are enough to make a difference to the extent consensus is assuming ($100b in increments NVDA rev vs $30b in incremental capex)