Every other company does this. All the time they overestimate

Yes. Gather data (including costs, actual usage, projected growth, and usage spikes), and write up a concrete proposal with well-thought-out alternatives (including keeping the status quo). Include engineering time when comparing costs.

Make it a discussion of tradeoffs, and go in willing to compromise and commit.