This question is actually exercise (20.5) from Set Theory by Thomas Jech. The original statement is:
Let $\lambda \ge \kappa$ and let $U$ be a normal measure on $\mathcal{P}_\kappa({\lambda})$. The ultraproduct $\mathrm{Ult}_U\{(V_{\lambda_x}, \in): x \in \mathcal{P}_\kappa({\lambda})\}$ is isomorphic to $(V_\lambda, \in)$.
Here, $\lambda_x$ is the order type of $x$.
I was able to shown (if I'm not mistaken) that if $M$ is the transitive collapse of $\mathrm{Ult}_U(V)$, and $N$ is the transitive collapse of $\mathrm{Ult}_U\{V_{\lambda_x}: \ x \in \mathcal{P}_\kappa({\lambda})\}$, then $N = V_\lambda\cap M$.
I have done this using a rank argument: Let $a \in M$ such that $\mathrm{rk}(a)<\lambda$, and let $f$ be the function on $\mathcal{P}_\kappa(\lambda)$ that represents $a$ in $\mathrm{Ult}_U(V)$ (i.e. $\pi ([f]_U) = a$ where $\pi$ is the collapsing function). Then as $x \mapsto \lambda_x$ represents $\lambda$, we have $\mathrm{rk}(f(x)) < \lambda_x$ for almost all $x$. So we may assume that this holds for all $x$, and then $[f]_U \in \mathrm{Ult}_U\{V_{\lambda_x}: \ x \in \mathcal{P}_\kappa({\lambda})\}$.
But why would $V_\lambda^M = V_\lambda$ hold in general? Moreover, what about the fact that $\left|N\right| \le \kappa^{\lambda^{<\kappa}}$, couldn't $V_\lambda$ be much larger?
Best Answer
As pointed out in a comment by spaceisdarkgreen, an almost identical question has been posted on MO in the past: https://mathoverflow.net/q/44289
It has recived an answer by Joel David Hamkins: https://mathoverflow.net/a/44291
For completeness, here is the answer given by Hamkins: