Two of the largest defense raises of the last few weeks arrived with something novel: a debt tranche. On August 27, REGENT Craft, the seaglider maker in North Kingstown, Rhode Island, closed a $240 million Series B split evenly between equity and debt, with the loan half coming from Erebor Bank. A week earlier, the hypersonic startup Castelion paired $800 million of equity with a $250 million revolving credit facility. For a decade, defense tech was a software story funded almost entirely with equity. The money now building the factories looks more like project finance. In the same fortnight, a sensing roll-up called Lyntris had to trim its IPO to get public. Capital is flooding the top and the bottom of this market and stepping over the middle.
The Deep Dive: The Debt Tranche Is the Tell
Start with what REGENT actually did on August 27. The company raised $240 million, described plainly in its own release as “an equal mix of equity and debt.” Mare Liberum and AE Ventures, the venture arm of AE Industrial Partners, co-led the equity. The debt came from Erebor Bank, the new, OCC-chartered lender co-founded by Palmer Luckey. The round brings REGENT to $340 million raised to date and lands right as the company moves from prototypes into production at its new 255,000-square-foot plant at Quonset.
Read the structure, not the headline. A software company does not borrow $120 million. It has no factory to secure the loan against, no equipment to repossess, no purchase orders a bank can underwrite. REGENT does. It has a building, tooling, a US Marine Corps contract for its Viceroy platform, and an order book booked years out. Those are collateral. Debt shows up exactly when a company crosses from selling code to building things, because atoms can be financed in ways that atoms-free businesses cannot.
Castelion tells the same story at a different scale. Its Series C, led by JPMorgan, Andreessen Horowitz and Carlyle at a $13 billion valuation, carried a $250 million committed revolving credit facility alongside the $800 million of equity. Castelion is not a slide deck. It is trying to mass-produce hypersonic weapons, which means production lines, test ranges, and long-lead materials. The presence of committed bank credit in a Series C for a missile company would have been unthinkable three years ago. It is now a signal of seriousness.
Who is lending, and why it matters
Erebor was built specifically to bank the industrial base. It was chartered to lend to defense, manufacturing and hard-technology companies, and its application to the OCC proposed capping loans at roughly half of deposits, a deliberately conservative posture for a bank underwriting hard assets. When a lender like that steps into a seaglider round, it is making a claim: American defense manufacturing has become creditworthy. Factories, contracts and backlog are now things a regulated bank will lend against, not just things a venture fund will bet on.
This is a maturation marker. Equity funded the experimental phase of defense tech, the years when nobody knew whether a startup could win a program of record or build at scale. Debt funds the industrial phase, when the questions shift from “will this work” to “how fast can you build it and how do we finance the plant.” The arrival of credit does not replace venture capital. It sits on top of it, and it only becomes available once the venture bets have been proven out. It also does something subtler. It lets a founder scale production without selling more of the company, which protects the equity that the earliest believers already own. A debt tranche is a vote that the collateral is good and the business will be around to repay it.
The middle is where it gets hard
Now hold that against Lyntris. The sensing company, assembled by Trive Capital from Accelint and Vitesse, priced its IPO on August 18 below its expected range, cut the deal to 17 million shares from 24 million, and opened about 11 percent below its IPO price on its first day of trading as “LYNX,” for a valuation near $1.8 billion on roughly $451 million of trailing revenue. This is a real defense business with real revenue, and the public market still made it work for the money.
Put the three side by side and the shape of the market appears. Enormous private rounds at the top, now large enough to carry their own debt. A busy seed and early-stage layer at the bottom. And a thin, unforgiving public window in the middle, where even a company with $451 million of revenue gets a haircut. This is the barbell we have written about before, and the debt tranche is the newest evidence for it. Money is happy to fund the giants and happy to fund the experiments. It is skittish about the companies stuck in between, the ones too big for a clean venture story and too small or too early for a warm public reception.
What the barbell rewards
The barbell also explains who the strategic investors are. Lockheed Martin Ventures sits in REGENT’s cap table alongside Founders Fund and Japan Airlines. The primes are increasingly acting like venture investors, taking early positions in the companies whose technology they will later buy or integrate, precisely as their own acquisition budgets tighten. When a company reaches the scale REGENT is now approaching, that early strategic relationship is what turns a demonstration into a contract, and a contract into the backlog that a bank will finance. The flywheel runs prime interest to venture equity to federal contract to bank debt to factory.
The Investment Takeaway
We read the arrival of debt as a bullish signal about the category and a clarifying one about stage. Debt is confirmation that defense hardware is real, that these companies have assets and revenue worth lending against, and that the industrial base is being rebuilt with private capital. But debt follows contracts and collateral, which means it rewards the late stage. It is not where a seed fund plays, and it is not what a seed fund provides.
For New North Ventures, we remain consistent about how we underwrite. We invest at the beginning, before there is a factory or a Marine Corps contract or a loan. The new question is whether a company is building toward atoms or software that will one day be collateral. How can the roadmap end in something a bank will finance: a production line, a fielded system, a backlog? The companies that clear that bar are the ones the barbell eventually pulls upward. We would rather own them at the seed, at a rational price, than chase them once the debt markets have already blessed them.
The Funding Ledger
A quieter back half of the fortnight for defense after Castelion’s billion-dollar week. Crunchbase called it a “sparser” lineup, and AI assistants retook the top of the table. The dual-use money that did move went to hardware and the infrastructure beneath it, and it spanned the full size range.
REGENT Craft, $240 million Series B (equal equity and debt), co-led by Mare Liberum and AE Ventures, debt from Erebor Bank. Seagliders for maritime defense and commercial routes. North Kingstown, Rhode Island.
Castelion, $800 million equity plus a $250 million credit facility, led by JPMorgan, Andreessen Horowitz and Carlyle at a $13 billion valuation. Low-cost hypersonic strike weapons. Torrance, California.
Muon Space, $250 million Series C, led by Eclipse. Satellite constellations for weather and intelligence, with a new San Jose line targeting up to 500 satellites a year.
Gatik, $200 million Series D, led by the Qatar Investment Authority and Koch Disruptive Technologies. Autonomous middle-mile trucking, a dual-use logistics play.
Emerald AI, $150 million Series A, led by Energize Capital and DCVC. Software that balances AI compute against available grid power, the energy layer under every defense data center.
Mara, $7 million pre-seed, led by Khosla Ventures with a16z Speedrun. Spike, a low-cost automated counter-drone system aimed at the cheap FPV threats reshaping the battlefield.
The pattern: after two weeks of megadeals, the biggest checks rotated back to AI, and defense money concentrated on things that get built. Counter-drone at $7 million and seagliders at $240 million are the same bet placed at opposite ends of the size curve. Investors want hardware and the picks and shovels around it, not more software.
The Northeast Desk
New Hampshire. Appledore Marine Engineering of Portsmouth won a firm-fixed-price contract for architect-engineering services inspecting Navy waterfront facilities in the Mid-Atlantic, an indefinite-delivery award from NAVFAC with a ceiling of up to $15 million and work running through August 2031. It is not a headline number, but it is exactly the kind of quiet, recurring federal work that keeps a small Seacoast engineering shop busy for five years.
Rhode Island. REGENT’s raise is also a jobs story. The company’s Quonset plant is expected to create 300 jobs, with the potential for up to 750 over a decade, and REGENT is eligible for up to $13 million under Rhode Island’s Qualified Jobs incentive if it hits its hiring targets. REGENT calls the $240 million round the largest venture raise in state history, and Providence Business News ran it as a record. A seaglider factory on Narragansett Bay, staffed by New Englanders, financed in part by a bank betting on American manufacturing, is about as clean a version of the dual-use thesis as the region produces.
The through-line for the Northeast is the one REGENT embodies: the maritime industrial base is where New England’s defense advantage compounds. The shipyards, the composites shops, the Navy relationships and the trained workforce are already here. Capital is now arriving to match them.
On the calendar. SENEDIA’s Defense Innovation Days ran this week at the Newport Marriott, August 31 through September 2, gathering national security leaders, industry and the next-generation workforce in the heart of southern New England’s submarine and undersea cluster. Watch the workforce sessions in particular. The hiring surge across the regional maritime base is the constraint that everything else now depends on.
Coming up. On Wednesday, September 16, BENS is convening leaders from across New Hampshire’s national security ecosystem, including industry, academia, and federal and state government, for a breakfast in southern New Hampshire on opportunities to strengthen the state’s role in America’s national security and economic competitiveness. The conversation will focus on defense innovation, supply chain resilience, workforce, advanced manufacturing, dual-use technology, and opportunities to grow New Hampshire’s defense industrial base. It is a good one for New Hampshire and northern New England, and New North Ventures is helping to host. Space is limited; contact jeremy@newnorthventures.com for details.
Portfolio Corner: Holocron
New North Ventures led the pre-seed round in Holocron Technologies, whose whole premise is that the best intelligence advantage now comes from public information, if you can read it fast enough.
Holocron, founded in 2022, builds an AI platform for what it calls global domain awareness in science and technology. It ingests open-source signals, research publications, patents, news, financial disclosures, and surfaces where foreign technology is moving before it becomes a classified surprise. The founding team is built for exactly this: Tristan Yang came out of US Special Operations Command, and Addam Jensen spent time at Intel and Sandia National Laboratories. They are people who have felt the cost of a technology surprise and set out to build the early-warning system for it.
The reason we backed it is the reason this newsletter exists. The center of gravity in intelligence is shifting toward open sources. A breakthrough in a Chinese lab, a quiet hire, a new supplier relationship, a patent filing, these are visible in public data long before they show up in a briefing. The hard part is not access. It is the machine reading and the pattern recognition at scale, which is precisely the problem Holocron is built to solve. It is the same discipline we practice every week here, turned into a product.
Holocron has kept a low public profile since the raise, and we are not going to invent milestones it has not announced. The question we are watching is conversion: whether a strong open-source analytics capability turns pilot interest inside the national security community into a durable program. The category is right and the timing is right. Execution is the variable.
More links to explore
Crunchbase’s weekly funding roundup for the week of August 22 to 28, on the “sparser” megadeal lineup.
REGENT Craft’s own release on the $240 million Series B and the move into full-scale manufacturing.
Lyntris begins trading on the NYSE as “LYNX,” the fortnight’s test of the public window.
Defense Daily on Mara’s $7 million pre-seed for mass-deployable counter-drone interceptors.