This summer's earnings calls sounded like a family argument over an expensive purchase. Ameriprise's chief executive, Jim Cracchiolo, called some competitors' transition offers unreasonable. Stifel's chief executive, Ron Kruskewski, put the question in its starkest form: either the biggest wealth firms are spending heavily on a business that may not exist in its current form, or the industry is about to be rebuilt around advisors who do more with AI. Recruiting deals now reach four to five times a recruited advisor's prior-year revenue.
The argument is being held over the wrong financial statement. Everyone is debating whether the next recruiting check is worth writing. But the industry has already written the checks: roughly $17 billion of advisor loans sits on the balance sheets of the eight firms that disclose them, advanced on the assumption that the funded advisors will produce the revenue that justifies the price. The question of whether AI breaks the economics of recruiting will not be answered on the expense line the executives are arguing about. It will be answered on the loan book, by an accounting rule that forces every firm to forecast the future of that book every quarter.
A recruiting check is a loan, and the loan is a claim on the future
Recruiting bonuses are mostly forgivable loans. The advisor does not repay the money if they stay with the firm and keep producing for seven to twelve years, the standard forgiveness windows in American Banker's reporting on the loan books. Stifel's own quarterly filing describes transition pay to advisors as forgiven through a charge to compensation expense over a five- to ten-year period tied to continued employment and performance. Until forgiveness, the loan sits on the firm's balance sheet at amortized cost, collateralized by production that has not yet happened.
A recruiting deal is therefore not a purchase of the past. The past is already banked; the loan is a claim on future output, the only thing that can repay it. When an executive says the firm may wait eight years to recoup its money on a cash basis, that is an admission about the duration of the claim, not just its size.
The size is worth staring at. LPL Financial's recruiting loans reached $3.68 billion in 2025, up 71 percent in a year and more than 1,300 percent since 2018, a surge tied mostly to retaining Commonwealth Financial Network's advisors after buying that firm. Morgan Stanley's book is the industry's largest at $4.86 billion at the end of 2025, up $520 million in a single year. Ameriprise holds $1.67 billion, up 200 percent since 2018. Raymond James holds $1.67 billion, and Stifel holds $745 million. Add Wells Fargo, UBS, and Merrill, and the disclosed total runs to about $17 billion. Wirehouse offers have roughly doubled as a multiple of trailing revenue in less than a decade, to 300 percent and beyond.
The allowance is where the future gets priced
Since 2020, the accounting standard known as CECL, short for current expected credit losses, has required firms to hold an allowance for expected credit losses on advisor loans, measured on the basis of historical experience, current conditions, and reasonable and supportable forecasts of what comes next. Stifel's most recent quarterly filing shows how thin those allowances run: roughly $33 million against advisor loans of about $745 million, a few cents of provision for every dollar lent. The rest of the industry's books are carried similarly.
The thinness is not negligence. Historical experience on these loans has been benign, because advisors mostly repay by staying. CECL takes three inputs, and the first two are quiet right now. The third is where AI enters. The rule does not let a firm wait until loans go bad to admit the future will differ from the past. The forecast must be baked into the allowance today, quarter by quarter, in the form of a number an auditor can argue with.
That makes the allowance the first instrument in the industry to absorb an AI thesis about advisor production. If the market concludes that AI will compress revenue per advisor within the life of these loans, that conclusion cannot sit in a presentation; it has to move an allowance, a charge-off line, impairment marks, long before it settles the argument over the next recruiting check. The executives debating recruiting costs are arguing about the income statement. The balance sheet is where the future gets priced.
The loans outlive the technology clock
The actual risk is a mismatch of horizons, a duration problem. The loans forgive over seven to twelve years, and the cash payback horizon executives cite runs to eight. The technology that could change per-advisor production is arriving on a shorter clock, and the firms' own roadmaps prove it.
Morgan Stanley was the first wealth manager to partner with OpenAI, and its assistant for advisors went live in 2023. Chief executive Ted Pick told investors the tool could save each advisor ten to fifteen hours a week, calling that "potentially really game-changing". By July 2025 the firm was pushing the tools to every employee in the wealth division, not just advisors. In February 2026, at a UBS conference, Morgan Stanley's wealth chief, Jed Finn, described the next stage: agents that "move money, open accounts, change beneficiaries," an AI that can interact directly with clients, and a portfolio construction engine, with the build-out set to continue through 2026 and beyond. Morgan Stanley frames the trajectory as enhancing advisors, not replacing them. The point is the timing: capability is arriving inside the life of loans written today.
A loan written in 2026 is priced against trailing revenue, a number from the past. It is repaid from future production, a number the lender is actively working to change. Those two diverge precisely when technology changes what an advisor can produce. Jason Diamond, who runs the recruiting firm Diamond Consultants, makes the same point in the language of deals: the packages assume advisors will hit growth and revenue-maintenance targets, and a world in which AI holds revenue flat breaks that arithmetic. The loan book is where that arithmetic lives.
Both sides of the debate assume the scarcity holds
Both sides of this argument have real cases, and both deserve their strongest form. The bears point at the obvious: if clients can get answers without an advisor, the revenue these loans are priced against may never materialize. LPL has acknowledged reviewing its cash-sweep policies, a lucrative use of clients' idle cash the firm has flagged as exposed in an AI world. The bear case does not even need AI to be right about everything; it only needs production to grow more slowly than the deal prices assume.
The bulls make the stronger case on the existing book. Phil Waxelbaum, whose firm Masada Consulting advises on advisor transitions, argues the big spenders are "playing the long game, not the short game": AI will make advisors more productive, which means the loans get repaid, and firms that pass now will find the best production spoken for. Raymond James's chief executive, Paul Shoukry, argues that recruiting cannot be switched on and off quarter to quarter and that firms win with culture and capabilities, because without a differentiated platform, the size of the check is the only selling point left. The strongest evidence for the bulls is Morgan Stanley itself: the industry's largest loan book sits under a wealth unit that just posted a record $8.9 billion quarter. The loans have not broken the balance sheet; they are funding it. Will Metzner, an advisor in Stifel's independent channel, makes the human case that "only a human can coach someone through a death, divorce, or retirement."
What the two sides share is an assumption about scarcity, and it is the weakest link in both arguments. Recruiting costs are high because productive advisors are scarce. The profession is aging: the average advisor is in the mid-fifties, Cerulli Associates has projected that about four in ten advisors will retire within the next decade, controlling roughly 40 percent of industry assets, and McKinsey projects a shortfall of around 100,000 advisors by 2034 if productivity stays where it is today. The loan is priced against that scarcity.
But the phrase about productivity staying where it is today is doing enormous work, and the industry's own AI investments are the main thing that could change the answer. Here is the contradiction inside the bull case: the same technology that makes the loans repayable, by raising what one advisor can serve, also makes the scarcity they are priced against less binding. If one advisor can do the work of one and a half, the industry needs fewer advisors, and the value of the production being recruited falls at the margin. The bulls assume productivity rises while scarcity holds. The bears assume production collapses while the loans stay collectible. Both cannot be fully right, and this analysis takes no position on which gives first. The point is that both cases trade in the same instrument, and neither side is watching it.
The balance sheet will keep score
The recruiting debate is really an argument about the price of human production, and the industry has marked that price with its own money. The loans are already on the books. The allowance must be re-forecast every quarter under a rule that demands judgment about the future. If the bull case holds, the allowance stays thin, the loans quietly amortize, and the argument settles in the bulls' favor. If the bear case holds, the first visible damage will not be a recruiting line on the income statement; it will be the mark on the loan book, in charge-offs and allowances, appearing years before the expense debate resolves. Either way, the balance sheet answers first, and it answers in dollars rather than opinions. Whatever you believe about AI and advice, the firms on both sides of the recruiting argument should want the loan book watched, because it is the only place where their disagreement becomes a number.
Primary sources
- American Banker reporter Dan Shaw's August 19, 2026 article for the executives' arguments, the four-to-five-times deal multiple, the 2024 to 2025 loan-balance changes, the Raymond James spending and production figures, the Morgan Stanley record quarter, and the cash-sweep review.
- Shaw's earlier May 19, 2026 report for the loan-balance history, the seven-to-twelve-year forgiveness windows, the deal-multiple escalation, and the UBS and Wells Fargo positions.
- Stifel Financial's quarterly filing on SEC EDGAR for the forgiveness-period and allowance accounting language.
- InvestmentNews for the Morgan Stanley AI rollout details, the advisor-shortage research, and the Metzner quote; Reuters coverage of Ted Pick's comments for the ten-to-fifteen-hour estimate; Financial Planning's report from the UBS conference for Jed Finn's agent roadmap; and Cerulli Associates for the retirement-wave projection.