a fixed effects (within) model does not have a single intercept: it has
N. I suggest you have another look at FE models' theory.
Some confusion often stems from Stata (misleadingly, IMHO) reporting an
"intercept" which is actually the average of the individual intercepts,
which you can recover in R as 'mean(fixef(<yourmodel>))'. Remember,
though, that this isn't "the intercept" of the estimated model in the
usual sense of the term. If you "need" an intercept, then you must
review your specification.