NextFin News - JPMorgan is relaxing its approach to lending against private shares, targeting founders and early employees of artificial-intelligence companies whose paper fortunes have swelled but remain locked up ahead of a public listing. The move is the latest escalation in a Wall Street race to win the next generation of ultra-wealthy clients, and it lays bare how much of the AI boom's wealth creation still depends on banks willing to lend against assets that carry no public price.
At JPMorgan Chase, demand for private-bank lending globally has surged tenfold in recent months, while at Goldman Sachs private wealth management in San Francisco loan balances have climbed 50% since 2023, according to interviews with bankers, lawyers and wealth managers. The collateral behind those loans is increasingly the same thing: private stock in AI and technology companies whose valuations have reached levels previously reserved for public megacaps.
The timing is not accidental. The US initial-public-offering market is delivering its strongest first half since 2021: traditional IPOs raised roughly $114 billion through June 30, more than seven times the $14.8 billion raised over the same period a year earlier, according to PwC's US Capital Markets Watch. Through the first six months, 65 traditional IPOs priced, compared with 34 in the first half of 2025. By mid-July, Renaissance Capital put 2026 proceeds at $141.2 billion, within striking distance of 2021's record $142.4 billion. The year's largest offering was SpaceX, which priced 555.6 million shares at $135 each on June 12, raising $75 billion at a $1.75 trillion valuation, the largest IPO in history by a wide margin and more than double Saudi Aramco's $29.4 billion debut in 2019. Anthropic confidentially filed in June, with investors pricing a potential listing at $1 trillion or more, and OpenAI filed confidentially in late May, eyeing a debut later in 2026. For founders and early employees, that creates a familiar problem at an unfamiliar scale: nine- and ten-figure net worth on paper, with mortgages, tax bills and expiring stock options demanding cash today.
"These are booming environments with a ton of activity, big IPOs, big index rebalancing, a lot of activity in Asia," JPMorgan chief financial officer Jeremy Barnum told reporters after the bank's latest quarterly results. "A lot of it is downstream of the AI theme, writ large on a global basis. It's just a very, very, very active environment."
Revenue at JPMorgan rose 27% to $58 billion in the quarter, with equities trading up 86% to $6 billion, a reminder that the AI boom is producing winners far beyond Silicon Valley. The bank's private wealth arm, part of an Asset and Wealth Management division run by Mary Callahan Erdoes that oversees more than $7 trillion in client assets, is now positioning to capture the next leg of those flows.
The Playbook: Lend Today, Capture the Mandate Tomorrow
Securities-backed lending to pre-IPO founders is not new. Major banks have offered such facilities for years, letting entrepreneurs borrow against illiquid shares rather than sell them and dilute their stakes. What has changed is the scale and concentration of the exposure, and the aggressiveness with which banks are competing for it.
JPMorgan's own product terms illustrate the mechanics. Its SpaceX share-backed facility is structured as a revolving line of credit with monthly, interest-only payments, a $2 million minimum, a requirement of at least 10 times coverage, an interest rate of SOFR plus 2 percentage points, and a 12-month tenor. The bank's securities-based lending disclosure is blunt about the risk on both sides: "A decline in the value of securities pledged as collateral may require you to provide additional collateral and/or pay down your loan," and JPMorgan can change its loan values "at any time, and without prior notice."
That coverage requirement is the key risk lever. Ten times coverage on a $2 million line means at least $20 million of stock must be pledged; on a $100 million facility, $1 billion of paper. Against a diversified public portfolio, banks commonly advance 50% to 70% of the collateral value; against a single private company's shares, the advance rate is far lower, because there is no exchange on which to sell the collateral if the borrower defaults. The discount is the price of illiquidity.
The economics of the trade are straightforward. The loan itself is rarely the profit center. It is a relationship wedge: a founder who borrows against pre-IPO stock today is a prime candidate for the IPO underwriting mandate, the post-listing wealth management account, and the family-office services that follow. Morgan Stanley, for example, reported more than $70 billion in net new assets from IPOs in the second quarter of 2026 alone, driven primarily by SpaceX. Winning the lending relationship early is how a bank gets first call on that wealth when it becomes liquid.
For the borrower, the appeal is equally clear. "Tax-aware borrowing strategies that take into account the after-tax cost of interest may help reduce the overall cost of accessing liquidity," JPMorgan's private bank advises clients approaching a listing. For a founder facing capital-gains tax on a sale, borrowing against shares can be materially cheaper than selling them. The math makes leverage look cheap, which is precisely how leverage becomes seductive.
The Risk Banks Are Underwriting
The catch is that this collateral cannot be marked to market every day. There is no public exchange, no continuous price discovery, and limited liquidity if a bank needs to sell. A 409A valuation is a backward-looking accounting exercise, not a market price, and tender offers happen infrequently and in small size. If the IPO window closes, those loans become harder to value and harder to recover.
Wall Street has lived through this before. In 2022, when technology valuations collapsed, share-backed loans to founders became problematic across multiple institutions. Founders who had borrowed against peak private valuations found themselves owing more than their shares were worth, with no public market in which to refinance. IPO volumes in 2022, 2023 and 2024 were among the thinnest on record, according to University of Florida IPO statistics, before the 2026 reopening. The difference now is concentration: today's pool of borrowers is disproportionately tied to AI companies whose private valuations in some cases exceed those of established public technology companies with comparable or greater revenue.
That is why JPMorgan's easing on one side of its balance sheet sits uneasily beside its tightening on the other. In March, the bank marked down loans to software companies held in the financing portfolios of private-credit clients, curbing how much those funds could borrow against them. "JPMorgan is being more prudent in lending against software assets," chief executive Jamie Dimon told investors at the bank's leveraged finance conference. One person briefed on the decision said the valuation haircuts did not trigger margin calls but were taken to pre-emptively shrink lending lines. The move was among the first signs of stress in the fast-growing private-credit industry, where funds borrow against loan portfolios to amplify returns.
The same collateral logic applies whether the borrower is a private-credit fund or an AI founder: when the asset has no public price, the bank's own model is the only discipline. Easing haircuts while the window is open and tightening them when it is not is rational risk management, but it also means the system is most leveraged precisely when the collateral is most vulnerable.
The infrastructure supporting these transactions is scaling fast. UK law firm Addleshaw Goddard reported that its stock-backed lending deals doubled between December 2025 and July 2026, a signal that the machinery is spreading across the financial system rather than sitting with a handful of private banks. Citi Private Bank, for its part, advertises $25.3 billion of liquidity provided globally through its margin-lending program against a broad range of assets including equities, bonds, ETFs and hedge funds.
Cyclical Euphoria, Structural Shift, or Both?
The right read is that both forces are at work, and they point in opposite directions. Getting this distinction right matters, because it determines whether the risk is a manageable cycle or a regime change that will not self-correct.
The structural leg is real. Relationship banking of this kind is a permanent feature of Wall Street: lend against the asset you cannot sell today in exchange for the right to manage the wealth that asset becomes tomorrow. The concentration of private wealth in technology is also durable; even if the AI bubble deflates, the next concentration will sit in another sector and the playbook will be the same. Securities-based lending itself is a standard product that predates the AI boom and will outlast it.
The cyclical leg is just as real, and it is what will determine who gets hurt. The current intensity of lending is tied to two transient conditions: AI valuations at record levels and an IPO window that has been shut for most of the period since 2021. Both are mean-reverting. IPO windows open on liquidity and close on it; the 2021 vintage of listings was followed by the weakest issuance years in decades. When the window closes, the borrowers who counted on listing to repay or refinance will be left with loans collateralized by paper that no one is bidding for.
The second-order risk is where this wealth migrates. Pre-IPO share lending effectively transfers the liquidity risk of private AI valuations onto bank balance sheets. As long as listings keep coming, the risk is invisible and the fees are real. If the window slams shut, banks are left holding collateral they cannot mark, cannot sell, and cannot easily restructure. That is a different risk from the 2022 episode, when the losses were borne mainly by founders and venture investors. This time, the banking system has taken the other side of the trade.
There is also a late-cycle tell worth watching. Competition for the next mega IPO is pushing banks to ease terms and shrink haircuts at exactly the moment when due diligence would argue for the opposite. In 2021, the same dynamic preceded a multi-year drought in new listings. Banks that lend most aggressively into a crowded window are usually the ones that learn the hardest lesson when it closes.
The Counter-Case: Why the Banks Think They Are Protected
The strongest argument against alarm is that these are not reckless loans. A 10-times coverage requirement, a 12-month tenor, and interest-only payments mean the bank is not lending into the equity; it is lending against a deep cushion of paper value. The facilities are also typically structured with personal recourse and cross-collateralization, so a founder cannot simply walk away. JPMorgan's disclosure that it can change loan values "without prior notice" gives it an escape hatch that did not exist in many 2022-era facilities.
Banks also argue that they are not underwriting the technology, only the liquidity event. The underwriting criteria focus on the probability and timing of an IPO, the depth of the cap table, and the quality of the lead investors, not on whether the AI model works. If a borrower's company fails to list, the bank expects to be repaid from a secondary sale, a refinancing, or the founder's other assets.
That defense holds in the base case. It breaks in the downside case, which is precisely why the counter-thesis deserves weight: the protection is only as good as the valuation it is measured against. A $1 billion paper stake at 10 times coverage supports a $100 million line only if the $1 billion is real. When private valuations are set by infrequent tenders and the entire sector is re-rated at once, the cushion can vanish faster than the bank can call for more collateral. Dimon's simultaneous decision to tighten private-credit lending suggests the bank's own risk officers know the difference between collateral with a price and collateral with a story.
What Comes Next
In the short term, the beneficiaries are clear: founders and early employees at AI companies get liquidity without dilution, and the banks that lend to them get first claim on the relationships that follow. JPMorgan is the most exposed of the large US banks to this trade by design, given the size of its private bank and its position in technology investment banking. A gauge of the five biggest US lenders — Bank of America, Citigroup, Goldman Sachs, JPMorgan and Morgan Stanley — had risen 14% so far this year through a recent close, ahead of the S&P 500's 8.3% gain, as investors price the banks as beneficiaries of the AI trade rather than just its lenders.
Over a medium-term horizon, the outcome depends on IPO execution. The base case is a steady flow of listings through 2026 and into 2027, with AI names leading but dispersion widening as investors scrutinize fundamentals. The upside case is an IPO wave that exceeds 2021, driven by a backlog that has only partially unwound. The downside case is a macro shock, a regulatory change, or simple market fatigue that closes the window while private valuations remain elevated.
The signal that would break the bullish read is specific and observable: if US IPO volumes excluding SPACs fall below roughly $100 billion for two consecutive quarters while AI private valuations keep rising, the "lend now, capture later" thesis is wrong. That combination would mean the exit door is closing while the collateral keeps appreciating on paper, leaving banks with loans they cannot realistically recover.
For now, the banks are not betting on the technology. They are betting on the liquidity event, and on being the ones who get to cash it. The risk is that when the event does not arrive, they are left holding the ticket.
Explore more exclusive insights at nextfin.ai.
