This is a risk question
dressed as a cost one
Build against buy gets argued on price, which is the least reliable part of it, because the price of building is the part nobody can estimate. The decision is easier when you ask what happens if it goes wrong.
The risks of hiring
Recruitment risk: the people who have run this before are scarce and mostly employed, and hiring the wrong senior person into a greenfield platform is a year you do not get back.
Key person risk: for the first year or two the platform lives in one or two people's heads. If they leave, the documentation will not be enough, because it never is.
Timeline risk: everything above happens before the first useful output, and it happens while the business is still expecting one.
The risks of buying
Dependency risk: you are relying on a vendor for something central. The mitigations are whether the substrate is open, whether the data is in formats you can read without them, and what you can still run on the day the relationship ends.
Fit risk: a platform built to be general may not do the specific thing your business needs, and you find out after committing.
Cost risk: the subscription is permanent, while a build is theoretically a one-off, though in practice the build has permanent operational cost too.
The question that resolves it
Is the data layer where your business competes? If your product is data, or your advantage genuinely comes from something in that layer, build it and carry the risks knowingly.
If the data layer is infrastructure supporting a business that competes elsewhere, buying is usually right, and the argument against is mostly about pride rather than economics.
The hybrid that actually works
Buy the foundation, hire for your data rather than for your infrastructure. One person who understands your business and your sources, working on a platform that already runs, produces more in a quarter than three people building from nothing.
This is also the version that leaves you least exposed, because that person's knowledge is about your domain, which is portable, rather than about a bespoke platform, which is not.
Common questions
What if we buy and later want to bring it in house? That should be possible without a rebuild, and it is worth checking before signing. The relevant questions are whether the storage format is open, whether the substrate is open source, and what you keep on the day you leave.
Does buying mean we never need data people? No, and anyone telling you otherwise is selling badly. You need people who understand your data. What you avoid is needing people who maintain infrastructure.