Nvidia could be a far larger company in a few years

This year, Nvidia expects the big five AI hyperscalers alone to spend nearly $800 billion on data center capital expenditures. That doesn’t include the many smaller players that are building their own AI infrastructure. I think it’s reasonable to estimate that total data center capex will be around $1 trillion this year, which means Huang’s projection points to the market tripling to quadrupling between now and 2030.

For argument’s sake, let’s say Nvidia can maintain its market share. I think this is a conservative estimate because a significant chunk of today’s AI capex is being spent on data center construction rather than on filling those facilities with chips and servers. Data centers take a while to construct, so the back half of the build-out will be more chip-loaded. This tilt would offset any potential loss of market share for Nvidia’s GPUs as other companies’ custom AI chips become more popular.

This year, Wall Street analysts estimate Nvidia will generate $412 billion in revenue. If Nvidia’s revenue can triple to quadruple based on these estimates, that would place its revenue between $1.2 trillion and $1.6 trillion annually.

Suppose Nvidia can deliver a 50% profit margin (its current level is 64%); that would indicate anywhere from $600 billion to $800 billion in profits. At a valuation of 30 times earnings, Nvidia would have a market cap of $18 trillion to $24 trillion; it’s about a $5.4 trillion company right now.

I’m not saying that Nvidia will deliver that kind of huge growth. Still, if the market it serves grows as Nvidia expects (and its position as a key supplier gives it a lot more information about the long-term build-out trajectories of its clients than investors have access to) and Nvidia maintains its market share, that would indicate the stock has huge upside ahead.