A lawsuit over a $400m yacht asks who owns the value when an AI erases the middleman

Nick Doron, CEO of Revolut, faces a lawsuit over his purchase of a $400 million yacht, according to reports on the case. The claim is that he used ChatGPT to identify the vessel's owner and then approached the seller directly, bypassing a brokerage commission of $20 million. These are allegations in a legal complaint; none of them has been tested in court, and no finding of wrongdoing exists at this stage.

Strip away the yacht and what remains is a clean illustration of where AI actually bites first. That $20 million was roughly 5% of the transaction price, and it was not payment for legal work, financing or logistics — it was payment for knowing who owned the thing and who might want it. Brokerage in high-value assets is an information-asymmetry business. Large language models are, among other things, very good at collapsing search costs. The commission did not disappear because a machine negotiated better; it disappeared because the scarcity the fee was priced on stopped being scarce.

The litigation will turn on something much narrower: whether a contract, a mandate or an agency relationship existed that entitled the intermediary to a fee regardless of how the buyer found the seller. That is a mundane question, and it is the same question that will decide thousands of similar disputes over the next few years — in real estate, recruitment, M&A advisory, insurance placement and every other trade where the fee is attached to a match rather than to the work of making the match.

Our reading: the headline invites schadenfreude about a billionaire dodging a bill, but the durable signal is about which parts of intermediation survive. Pure discovery — finding the counterparty — is the layer AI compresses fastest, and it is exactly the layer that carried the fattest margins. What does not compress is judgment, trust, access to people who will not answer a cold email, and accountability when a deal goes wrong. That is the same split we have tracked across law, banking and real estate: the routine informational middle thins out, the relational and expert edges hold. Expect contracts in these industries to be rewritten first — not to ban AI, but to attach fees to outcomes and mandates rather than to the act of introduction. Short term, that is a messy and litigious transition for a lot of people whose income depends on knowing something the client cannot easily look up. Longer term, cheap discovery is how markets get more efficient and how the value migrates toward work that actually requires a human.