A graph this big takes a team. Two senior data engineers. One identity domain expert. Twelve to eighteen months to a working v1. Ongoing operations after that — sources break, schemas drift, the resolution rules need tuning. We did this work. The cost was real and we paid it. The result is what your team can have on day one for a fraction of what it took us to ship. Build it yourself is the right answer when identity is the core competency you want to own as IP, when you have the engineering bench, and when 12–18 months of internal build is acceptable. We are the right answer when identity is infrastructure, not IP.
For most teams, identity is a substrate — a thing that has to work, a thing the rest of the business depends on, a thing that should be invisible when it works. That is the definition of infrastructure. Infrastructure is bought, not built. The companies that won by building their own identity (Stripe, Plaid, the rare exceptions) had identity AS the product. Most don't.