The cash spread is what the selloff repriced

LPL Financial and Charles Schwab each fell more than 6% on September 23, and the analyst notes that moved with the tape did not blame the advisory fee; they blamed the cash sweep, the spread a custodian keeps on client money parked between decisions. AI anxiety was the trigger and parked cash the destination, which makes a single bad Tuesday a repricing of the asset under the advice—the client cash a platform holds—that an advisory-fee multiple never sees.

Because the notes named a destination without drawing the map, the mechanism is worth spelling out. Uninvested cash in a sweep account pays a yield most clients never audit, and it sits still precisely because moving it costs a phone call, a form, and a reason. Drive that friction toward zero—an agent that compares the sweep yield against a money-market alternative and executes the transfer in one instruction—and the spread the custodian keeps gets bid away by arithmetic. That reading is inference, labeled as such, but it is the reading that fits where the notes put the threat.

The sell side does not usually reach for the sweep when it talks about AI; the ordinary fear is that software eats the advisor's charge and the human fee shrinks. Tuesday's notes inverted that, treating the advisory fee as the durable line and the cash spread as the fragile one. The concession matters: whatever AI does to advice, the charge has a contract behind it, while the spread is an arbitrage the client has never had a reason to close.

Three firms moved on the client record the same day. Choreo bought onboarding automation with a custodian pipe attached; Callan Family Office hired someone to own its data; Savvy Wealth raised money to keep selling software into a book it does not custody. None bought a fee; each bought a claim on the account the fee gets charged against.

The advisory fee is the revenue anyone can see—disclosed, benchmarked, multiplied—while the cash spread is the revenue that exists because clients are busy, and it has never needed a line item because nobody has had to defend it. An advisory-fee multiple prices a stream of annual charges on assets; it says nothing about the cash sitting in those assets between trades, a second revenue line riding on the same relationship and collected by whoever holds the account. Two platforms can post identical fee revenue and very different cash economics, and the equity market has not had a clean way to tell them apart. On September 23 it may have found one.

Choreo took the most literal route: the PE-owned acquirer bought Beemo Automation and with it a direct Schwab connection for onboarding throughput, and it is the connection rather than the automation that carries shelf life. Throughput bought at a monthly retainer is a service bill; a direct pipe into a custodian is an asset, because a client record lives inside the custodian's systems, and whoever holds the connection decides how fast an account and its cash can be opened, moved, and held.

Callan Family Office took the slow version, hiring a data chief at director level to own the client record and the reporting stack behind it. A salary is the cheapest switching cost a family office can buy, and it covers the half of the platform war a software contract cannot: the part where someone inside the firm, rather than a vendor, answers for the data. Choreo buys the pipe, Callan hires the person who points it—the same bet at two clock speeds.

Savvy Wealth took the fastest route and the shallowest one, closing a $100 million round while renting Fidelity's balance sheet. The arrangement is what it sounds like: a branded front end on top of a custodian's rails, with the account-level record still belonging to the clearing broker. That is the cleanest of the three structures and the most exposed, because Savvy owns the relationship and the software it sells into it while the account, the cash, and the spread sit one layer down, at Fidelity. A hundred million dollars buys speed, and the record stays where the clearing broker keeps it.

Set the three side by side and they rank by depth of ownership: Choreo is buying a claim on the custodian's systems, Callan is buying the person who keeps the record straight, and Savvy is renting the ledger it needs. Only one of those positions survives a client deciding to move the cash.

The most consequential line in the same week's news belongs to a firm that did not have to spend. PWD's tracking shows Schwab has turned Ameriprise recruiting into a serial launch program, with a fourth billion-dollar Ameriprise team due out in 2026—a cadence that prices the template rather than the team. The mechanics are dull and decisive: a recruited team brings its client record onto the custodian's platform, and the record is what the cash sits inside. A custodian can hold the asset without buying a single RIA; it only has to keep the team.

That is the asymmetry a wealth platform's deal model is slow to price. A buyer pays for the client relationship, the custodian keeps the account, and the sweep sits between them, which means an acquirer can spend a decade paying advisory-fee multiples and still lose the cash because the cash follows the record and the record follows the custodian. That is why Schwab's 6% deserves a second look from anyone who read it as a verdict on advice. The firm marked down hardest is the one holding the asset the notes were worried about.

Vanguard's move to buy the last pillar of custody revenue made the same point in a single transaction: with ticket charges gone and fund-company revenue sharing fading, the client record and the cash spread are what remain to own, and an asset manager now owns both. AssetMark's addition of two interval funds to a $91.8 billion platform belongs to the same logic from the other end of the shelf. On a wealth platform the wrapper and the client record decide whether an account stays at all, which is why curation keeps turning up wearing a product label.

None of this makes the September 23 selloff wrong; it makes it early, and imprecise. The bear case is simple enough to state so that it can be judged: if the AI threat to the sweep is real, then the earnings a custodian draws from client cash is a shrinking asset, and both LPL and Schwab were marked down correctly. If it is one more automation headline that never reaches the sweep, then the 6% was a chance to buy two established distribution networks cheaper. The two cases cannot both be right, and the deal market will settle the argument before the tape does.

The pipeline agrees. Deal volume for 2026 survives on mandates already signed, which pushes the hard negotiating into how the next 500 transactions get paid for. A buyer who has watched a custodian take the spread while the seller keeps the fee multiple has an obvious adjustment to make: stop paying a premium for fee revenue and start paying for the account. The mandate signed next will price the record, and the seller still quoting a fee multiple will be the last to find out.

The selloff found the right destination and the wrong sequence. The cash spread is the exposed revenue line, but exposure runs toward whoever rents the account and away from whoever owns it, which makes the day's 6% at Schwab look like the market marking a risk at the firm strongest against it. The cleaner trade is the one the week's deals already made: the record is a better asset than the fee, and the deal market has not repriced it yet. Choreo, Callan, and Savvy are each paying for the thing that survives fee compression—the pipe, the person, the relationship—while the RIA any of them might buy next is still valued on a multiple of a revenue line that a smarter agent can squeeze.

The repricing, when it comes, will not arrive as a lower multiple. The next 500 RIA transactions will be settled in how buyers pay rather than what they pay, and the currency of a deal reveals what the buyer thinks the asset is worth. Two numbers will settle it before the deal tape does. The first is the sweep yield a custodian reports against the money-market alternative, the gap a client can act on. The second is the Ameriprise launch cadence on Schwab's platform, where the fourth billion-dollar team is due out in 2026. If the gap closes, the notes were early rather than wrong. If the teams keep arriving, the custodian is buying the record one recruitment at a time, and the fee is the last number anyone reprices.