The factory floor in Southern California runs about ten robots for every person on it, by its founder’s account, machining parts for aircraft, ships and weapons. Hadrian, the company that built it, now counts the largest bank in the United States among its shareholders, which is not where you would have looked for JPMorgan Chase a year ago.
On Monday, JPMorgan Chase said it had put 5 billion dollars of its own equity into companies like that one during the first year of its Security and Resiliency Initiative, half of the 10 billion dollars it set aside for startup investments and ahead of the schedule it published. The bank also said it had financed and facilitated about 230 billion dollars of transactions tied to the program, part of a 1.5 trillion dollar commitment it made over a decade.
The initiative was announced on October 13, 2025, and covers supply chains and manufacturing, defense and aerospace, energy, frontier technology, and pharmaceutical and health technology. It is run day to day by Todd Combs, the former Berkshire Hathaway executive who reports directly to Jamie Dimon, and overseen by co-President Doug Petno, who told CNBC the bank may go past the original 10 billion dollars.
“There’s a lot of really interesting companies that have been working to solve some of these problems,” Petno said, describing early venture-stage manufacturers that have found ways to build capabilities at home rather than buy them abroad. Combs, in a statement, put it in the language of patient capital: “Our approach is to find generational companies and partner with them for the long term.”
Two of the holdings are public. Hadrian, which raised 1.4 billion dollars in August in a round the bank helped lead, sells automated factories to aerospace, defense and maritime customers. Castelion, founded in late 2022 by a group of former SpaceX engineers, raised 1 billion dollars that same month and is developing a low-cost hypersonic missile called Blackbeard. Most of the other positions have not been named.
The distinction that matters is between the two numbers. Financed and facilitated means loans, bond underwritings and advisory work, most of it other people’s money collected at a spread. Equity means the bank owns the residual claim on a company that may not have a product, a customer of scale or a path to either. One is fee income with credit risk attached. The other is capital that can go to zero.
That is a departure for a firm that spent the past decade retreating from principal investing, partly because regulators and shareholders prefer banks that lend and advise. Equity stakes in private manufacturers carry capital charges, require quarterly marks, and are difficult to hedge. A bank can sell a loan. It cannot easily sell a factory that has not been built.
The timing explains the urgency. Dimon framed the program around the wars in Ukraine and the Middle East and around trade disputes with China, arguing that the components and minerals American industry depends on had become a strategic exposure rather than a procurement question. The announcement landed while the administration was pressing companies to move production home, and executives who praise that agenda are also, plainly, courting the government that regulates them. Both readings are available, and neither is wrong.
What the money is buying, mostly, is capacity. Hadrian’s model is a factory per customer segment, each one staffed thinly and tooled heavily, which is expensive before the first part ships. Castelion is trying to build a weapon at a fraction of the unit cost of the incumbents, a claim that requires production runs to prove. Neither business produces the kind of cash flow that services debt, which is precisely why they need equity and why a bank’s balance sheet is an unusual place to find it.
Petno’s explanation for the pace is that the opportunities were there. The skeptical version is that a finite number of credible teams can absorb 5 billion dollars in twelve months without the price of their equity rising to meet the money. Defense technology has been the best-funded corner of venture for three years, and the list of companies with a real product and a real contract is shorter than the list of companies with a deck.
There is also the question of how the bank gets out. A stake in a private manufacturer becomes cash through a sale or a listing, and the buyers for defense suppliers are a small set of primes that have their own consolidation agendas and their own view of what a program of record is worth. The customer, in most of these cases, is the same Pentagon that has to fund the program. Public markets have been hospitable to a handful of names and indifferent to the rest.
For the bank, the exposure is manageable in dollar terms. Five billion dollars is a fraction of a quarter’s earnings, and the positions are diversified across sectors. The reputational exposure is less tidy: a single failed factory or a missile that misses its cost target becomes a story about the bank that said it was rebuilding American industry, and the second year of a ten-year program is when the first write-downs usually arrive.
Dimon has described the effort as a bet on the country rather than a bet on any company. That framing is comfortable and also untestable, which is why the numbers in the second year will matter more than the ones in the first. A bank can decide in a quarter to finance a supply chain. Building one takes longer than the loan.
