"One gig uplink may well handle about 50 gig access port."
Typically, user ports have low overall average utilization. If they only average about 2% utilization, it would take 50 of them to push a same bandwidth uplink port to 100% utilization. However, "your mileage might vary".
In your OP you note you're moving to a new location. Ideally, you would look at existing (i.e. prior locations) user usage stats. Assuming, how they will continue to use a network doesn't much change, you can pretty accurately predict what you'll need for uplink bandwidth relative to the number of user ports.
Or, let me put it another way, if you're thinking you need a 40g uplink port. If your user ports are gig, that means you could support 40 gig user ports running, concurrently, at 100%, or 400 gig ports running, concurrently, at 10%, either is probably very unlikely.
Also keep in mind, it would probably be a good thing to have two uplink ports, for redundancy. Lots of ways to take advantage of those to effectively give you twice your single uplink bandwidth. So, if one gig uplink port might support 50 user edge gig ports, a second gig uplink port would now allow, possibly, supporting 100 user edge gig ports.
Years ago, I was at a company that was planning to deploy 48 port 3750Gs, stacked, for user edge devices. Some stacks would go the full stack limit of nine stack members, supporting 432 gig edge ports. Originally we proposed to only use two Etherchannel gig uplinks, with the option to increase the Etherchannel to 4 or 8 ports, if needed. Well, someone, I forget who, thought with that number of edge ports we had to deploy with an 8 gig port Etherchannel. So, we were going to do that, until when we worked the numbers, it was actually less expensive to swap out two of the 3750Gs for 3750Es, so we could use 10g uplinks (2). (What made the 8 gig port Etherchannel option more expensive was the 16 [Cisco] optical gig transceivers.)
So, we went ahead with the dual 10g uplinks, and even for 432 gig user edge ports, the 10g uplinks had pretty low utilization.
Again, your mileage might vary. Most user apps (at that time) had the app software on the host, so you were only opening then saving data files, or you were using some HTTP web app, which often don't move a lot of data either.
