Then pointwise convergence means that for each xx and ϵϵ you can find an NN such that (bla bla bla). Here the NN is allowed to depend both on xx and ϵϵ.
In uniform convergence the requirement is strengthened. Here for each ϵϵ you need to be able to find an NN such that (bla bla bla) for all xx in the domain of the function. In other words NN can depend on ϵϵ but not on xx.
The latter is a stronger condition, because if you have only pointwise convergence, it may be that some ϵϵ will require arbitrarily large NN for some xxs.
For example, the functions fn(x)=xnfn(x)=xn converge pointwise to the zero function on RR, but do not converge uniformly. For example, if we choose ϵ=1ϵ=1, then the convergence condition boils down to N>|x|N>|x|. For each x∈Rx∈R we can find such an NN easily, but there's no NN that works simultaneously for every xx.