How do i decide the compute sizing for a portal

Question

I am designing the government portal that is gonna provide scholarship to various students under numerous schemes from GOI. In that way around 40 million users would be there to use the portal. how should i come up the compute sizing for this like how many servers to be created, what ram, and cpu to be given.

it will have front end where user can from web as well as mobile app, which is gonna have a mobile module, then student registration, then their academic details, registeration details, aadahr api to check its detail fro authorization. once done instatant verification will start with api gateway for student database integration, with several nodal officers for caste, religion, disability certificates, with colleges, institues, education boards, school institutes and then once verified, go to payment and its api

Answer 1

Load Testing
You build the solution with the smallest/cheapest hardware you can and you work hard to ensure that it runs correctly on small hardware.

When you think you have something ready for production, you begin Load Testing.

To do that, you generate artificial user interactions and you monitor how your system behaves. If it works, you increase the number of user interactions until either something stops working or the number of interactions is at least double your expected production load.

If it doesn't work, you change one parameter at a time (memory size, number of cores, number of servers, allotted bandwidth, etc.) until you figure out where your limits are.

Done right, it's a long, tedious, and potentially expensive operation that is absolutely necessary for any solution that is likely to see more than a few users at a time.