Post 19: If Arch is decentralized, why does it keep naming four companies? Here is the question I kept dodging, so let me ask it out loud: if @Arch is supposed to be decentralized, why does every serious explanation of it end up naming four companies? Anchorage. Fordefi. Gauntlet. Chaos Labs. If you have read my posts for a while, you have seen those names. They are not decoration, and they are not there to impress you. Each one does a specific job, and understanding the jobs is the fastest way to understand what Arch Prime actually is. Anchorage and Fordefi do custody. That is the unglamorous work of holding keys the institutional way, with controls and segregation and audit trails, so that a prime brokerage client is not trusting someone's laptop. Gauntlet does risk: the models behind liquidation rules and loan safety, the math that decides when a position gets closed and how. Chaos Labs does oracles: the price feeds that programs read before they move value, because a lending program that does not know the price of bitcoin is just guessing. Now the uncomfortable part. Naming companies sounds like centralization, and if you stopped there, you would be right to be suspicious. Here is the distinction that resolves it, and it is the most important distinction in the whole Arch story. Arch Network is the neutral rails: its own execution and settlement network, its own validator set, its own virtual machine. Arch Prime is the business built on top of it, the anchor tenant. And a prime brokerage, in traditional finance, IS its counterparties. That is the product. When a hedge fund picks a prime broker, it is picking the custody, the risk desk, and the clearing. So of course Prime names its partners. Hiding them would be the red flag. Here is what those companies do not do. They do not vote in consensus. They do not sign your bitcoin. That job belongs to the validators, who reach 51% plus consensus and then threshold sign the native Bitcoin transaction with FROST plus ROAST. No single party moves funds, partner or otherwise. The honest version of the risk: if a partner failed, Prime's products would feel it. The network's signing rule would not change. And the real trust questions stay exactly where they have always been: an honest validator threshold, correct code, and a correct indexer. So judge each layer by its own test. The network by its signing rule. The business by its counterparties. Mixing the two up is how you end up either trusting too much or dismissing something real.