To start, I assume by client you mean a website frontend, and by server you mean the backend for the website. (It can get a little confusing, since there is also a WalletConnect client which talks to the wallet.)
The WalletConnect client acts as a gateway to communicating to the user’s wallet. And this gateway tends to work best when placed in a frontend that the user is also interacting with, since they’ll be able to see both sides of the interaction – the website frontend and their wallet.
I’m sure you could figure out a way to establish a WalletConnect session in the browser, then transfer that session to a backend server somehow, though it would be more involved than just copying the clientId, since the WalletConnect client and wallet establish a secret key used to encrypt messages. You would need to extract at least this secret key information (probably from local storage, but I don’t know exactly where it resides), shuttle it to your backend, and somehow use it to bootstrap another WalletConnect client.
While the above is theoretically possible, I don’t see how it increases security. The same information is still going over the WalletConnect bridge server and being analyzed by the wallet. And this information only consists of a list of unsigned transactions, which will be public as soon as you submit them to the chain anyway. Though perhaps you are using some private/secure information to determine the list of transactions that should be signed? In that case you could always compute the unsigned transactions on your backend, then send them to the frontend and have the frontend send them to WalletConnect.
You’d need to share more about the threat model you are trying to protect against for me to evaluate whether or not there’s a meaningful security benefit to offloading the WalletConnect interactions to a backend server.