Happy Tuesday, Team PA. Every so often, if you know where to look, you’ll find a government artifact that cuts through the noise and offers remarkable wisdom. One landed on our desk this week, courtesy of the Pentagon’s Manufacturing Technology office (ManTech for short). And then the Army followed it yesterday with something it says it has never done before. Saddle up! 🤠
IN THIS WEEK’S EDITION:
🏭 Our suppliers can’t surge
🤠 Enter the Rodeo: the Army’s load test
🧾 Stats: 7-to-1 private: federal; the marine frontier; & tungsten’s comeback
⚛️ Hypersonic triple-header, GOES, uprating, & Sparrows Point
🌵 Austin’s first Deep Tech 50
Forwarded this? Join Per Aspera today.
ManTech opened in 1956 to fund new ways of making things rather than the things themselves. On Aug. 10, it had a 70th birthday party at the Pentagon, where Under Secretary Emil Michael called ManTech “the unseen force” behind precision-guided munitions, stealth aircraft, and advanced microelectronics.
Seven weeks later, ManTech published its plan for the next five years. The office did not soften its view, writing on Page 4: “Today, we lack the latent capacity to rapidly scale production of key systems and munitions in a crisis… Our inability to surge production is a known vulnerability that invites aggression.”
Munitions & missiles are perhaps where the production pain is most publicly visible today, given the Iran campaign’s drawdown of prewar stocks in key categories.
But we face plenty of other acute production challenges: the plan names castings, forgings, and microelectronics as single points of failure (SPOFs, as our friends @ Overmatch call ‘em).
The Pentagon’s new defense manufacturing plan recognizes a fundamental truth: military superiority is tied to production capacity. America can deter and defend against military conflicts by creating and producing the world’s most modern offensive and defensive weapons systems. The key is rebuilding our industrial capacity to manufacture faster, better, cheaper, and at overwhelming scale.
How did we get here?
Well how much time do you have? The plan identifies three habits from the long peace — the 'end of history' that never arrived:
Just-in-time. Under this paradigm, inventory becomes a sin. “Slack” is a cost center to be engineered out of the system, but that same slack is what you call on in a time of crisis (or extreme demand) to surge.
Consolidation. At the Last Supper in 1993, the Pentagon famously told its primes: merge, or die.
Offshoring. “Critical production” is as specific as the plan gets about what left, and “abroad” is as specific as it gets about where it went. We can fill in between the lines: the ‘schlep’ work (dirty, low-margin, capital-heavy) followed the lowest bid out of the country first, but as went foundries, so too did forge shops, then fabs, and more.
DIB consolidation: In 1990, 13 American companies built tactical missiles. By 2020, only three did. Many such cases!
When the Pentagon doubles an order, nothing much happens at first. Spinning up qualified tooling, spare machines, trained operators, and a second source can take years. It can feel brutally slow to the new “deep tech” investors who have joined our ranks in search of a durable investment thesis, but are acclimated to the breakneck rate of change in AI SI capabilities.
ManTech’s answer to this “brittle” system is to distribute the work: spread production to a deeper bench of suppliers, and lean on additive manufacturing, robotics, cold-spray repair, so no single plant can stall a program.
Same part, every time
There’s always a catch, and ManTech takes this one on candidly: a qualified part is a qualified process. If you change the machine, or the foundry’s cooling rate, you have a different part until testing says otherwise. In defense that means first-article inspection and “a burdensome cycle of extensive physical testing for every minor change.” This is why a second source can take years, and why an item manager can’t currently send a design to a certified cell at a depot and get a flight part back. ManTech’s proposed fix is to put the instruments inside the machine (in-situ monitoring, CT inspection) and maintain a “digital thread” tying how a part was made to how it performs, allowing qualification to run on data, targeted tests, and tight feedback loops.
All roads lead to a valley…
This takes us to the all-too-familiar standoff that prevails in the gap between research funding and a production contract. Investors want to see a customer; the customer wants to see a mature, de-risked product. The companies caught in between will often abandon good technology or exit to bigger players, producing a DIB “smaller, less competitive, and slower to respond than the national security environment demands.”
GAO counted 42,723 small businesses winning Pentagon contracts in 2011 and 24,296 in 2020, a 43% drop.
This is the all-too-(in)famous “Valley of Death,” which the plan has rebranded the “transition corridor.” ManTech can fund pilots and de-risk a process, but knows its limit, and draws that line: “this strategy tackles the challenge from a manufacturing technological standpoint.” The POs that end the standoff have to come from the acquisition system. And, to the Department’s credit, that system is being reformed and rebuilt, with haste, before our very eyes.
What we hadn’t quite figured out was a way to find out, before a war, which suppliers can actually deliver under load. On Monday the Army proposed one. 👇
We’ve all spilled enough ink on the diagnosis. Now, in the Army’s telling, it’s time for a load test.
On Monday at AUSA’s opening ceremony, Acting Army Secretary Adam Telle announced the first Manufacturing Rodeo, for early next year: “a race between our industrial base partners that strains our organic industrial base, stresses the supply chain, pressures manufacturing processes, and shows us exactly where it breaks.” His example was the one-way attack drone, which the Army wants by the million: “It may look simple, but it’s a flying supply chain we don’t control enough of.”
We like it for a few simple reasons.
Bias to action: Each assessment/indictment of our DIB that we’ve read was done on paper. The Rodeo proposes to find the breaking points by actually loading the thing up.
Competition: The best answers to our DIB problems will come out of competitive processes, not central planning.
High mix: Army depots are but one small corner of the DIB, but this test generalizes pretty well. We field hundreds of systems with thousands of part numbers ordered in small lots on an irregular cadence. We optimized for the opposite — a few primes running predictable low-mix throughput. Companies like Hadrian and Divergent (a PA Founding Sponsor), who built their factories around exactly this problem, are well-positioned for this!
Digitalization: “Digitalizing analogue processes” sounds like a line out of Davos, which is to say corporate slop, but it’s a cornerstone of any plan to turn the ship around. It’s the only way we can liberate and scale our tribal knowledge, it’s what makes a qualified process portable and auditable, and in this case, it’s how you score a rodeo.
…While we’re here: The DOW and SBA launched the Smaller War Plants Commission in August, with more details being shared here this week. Aaron Slodov notes some of the underappreciated changes for American manufacturers:
Up to $10m in combined SBA financing
Repeat 504 loans for distinct projects
Up to $5m in manufacturing revolving credit
90% loan guarantees
7-to-1. Bain and AIA’s new report, The State of Investment in Defense, says America has plenty of defense capital and too few places to put it: the top 100 venture-backed U.S. defense companies have raised $118B of private capital since 2020 against $16B of federal obligations, seven private dollars for every government one. The report’s list of where capital still can’t reach is a mirror image of the ManTech plan: castings, forgings, chips, and precision components.
$3B. That’s nearly the amount of venture that’s been put to work in marine startups in the last 12 months, per Crunchbase, with Saronic’s $1.75B taking more than half and a16z/Founders Fund each backing three of the 27 companies on the list. The ocean is nearly three-quarters of the planet, and the market is finally waking up to it. To us it looks a lot like the new space economy did a half-decade ago.
First concentrate since 1993. On Sept. 23, Almonty announced that its Sangdong mine in South Korea, once one of the largest tungsten mines in the world, has produced its first concentrate in 33 years. The same week, Maine’s Elmet agreed to pay ~$125M for 4.99% of Masan High-Tech, owner of Vietnam’s Nui Phao mine and nearly all of that country’s ~2,400 tons a year. More than 90% of Sangdong’s first phase is already contracted for 21 years to Global Tungsten & Powders in Towanda, PA, and on Jan. 1 the Pentagon’s ban on Chinese-sourced tungsten takes effect. China still refines most of the world’s tungsten, but Allied/Assured supply is being rebuilt one mine and one stake at a time. 💪
001 / HYPERSONIC TRIPLE-HEADER: Last week, the Air Force confirmed the first flight-test release of the Hypersonic Attack Cruise Missile (HACM), dropped from an Australian Super Hornet over a range Down Under, a milestone first promised for 2024. A few days later, Castelion posted a photo of Blackbeard firing from a truck-mounted launcher, our first look at the ground-launched variant. Finally, South Korea joined the Mach 5+ club on Oct. 8 with the first flight of a homegrown hypersonic glide vehicle. (While we’re here: a Navy-administered consortium picked Anduril, Lockheed Martin, and Ursa Major the same day for the first awards under its next-gen hypersonics prototype program.)
002 / Here GOES nothing ... As discussed last week (#065), some kinds of steel are getting hard to come by. Now the federal spotlight is on grain-oriented electrical steel (GOES), the magnetic steel at the core of every transformer. Last week, Energy Secretary Chris Wright toured Cleveland-Cliffs' Butler Works in western Pennsylvania, the sole U.S. producer. Cliffs is putting $200M into the mill, backed by a $75M DOE award, to lift GOES output by up to 25% with new induction reheat furnaces built in Warren, OH, with completion expected in early 2028. We’ve discussed transformer lead times time and again around here. (For the @BottleneckBros in the back: datacenters need power → power needs transformers → transformers need GOES → all U.S. supply comes from Butler, PA). We’re happy to see this ramp, and we’d be even happier to see a second source. 👀
003 / A NEW TERM TO LEARN: UPRATING... Roughly 1.5 GW of nuclear was financed over the course of about a week, all without a single new reactor! On Oct. 6, Constellation & Google signed for 890 MW of uprates across 11 units at six plants in PJM, backed by $4.3B+ of Constellation’s own money. This follows Amazon’s Sept. 30 20-year PPA at Calvert Cliffs, which pays for a $3B, 190 MW expansion of a plant that first opened in 1975, and DOE’s offer of up to $4.2B in loans to Vistra to add 433 MW at Beaver Valley, Davis-Besse, and Perry and run them 20 years past their current licenses.
Inelastic demand: In #046 we called hyperscalers price-insensitive buyers lining up for whatever nuclear they could get, given it’s a clean, stable baseload a 24/7 datacenter can sign for two decades. An uprate is the fastest version on offer. (Also: Constellation’s 890 MW pencils to under $5,000/kW, against ~$15,000 for Vogtle’s two new units.)
Right where it hurts: All of this lands in PJM, whose last capacity auction came up 6.8 GW short of its reliability requirement for the year starting June 2028. Hence the “all of the above” posture for industry and DOE, and renewed focus on uprating, as brownfield > greenfield for time to power!
004 / BACK TO SPARROWS POINT... On Oct. 6, Anduril and the Navy announced Arsenal-2, a $6.6B, 2M+ sq ft yard on the old Bethlehem Steel site in Baltimore County that will build Virginia-class parts, from torpedo tubes up to large sections like the bow, for Electric Boat and Newport News (adding ~15% more Virginia-class capacity). The Navy wants two boats a year by 2032. In #061 we approximated an equation for this: ships + repair capacity + trained labor + supplier capacity + production speed. Arsenal-2 takes aim at the last three.
The historical symbolism is strong with this one. Sparrows Point was once the largest steel mill in the world, and its yard delivered more than 100 ships between 1939 and 1946. We’ve done it before… we’ll do it again!!!!
001 / FILL ’ER UP... Orbit Fab has qualified the RAFTI HP, a 5,000-psi refueling port for the noble gases that electric thrusters use (xenon, krypton, and argon). RAFTI is one of just two refueling standards the Space Force has blessed. Refueling technology + satellite-servicing spacecraft (s/o Starfish Space & Otter, the first full-size one of which launched Oct. 1) could be a paradigm shift in which ‘single-use’ satellites get a new lease on life. You can imagine where this might pencil: higher orbits, higher-value satellites, and situations, as the Pentagon calls it, where “sustained maneuver” is indispensable.
002 / POSITION, NAVIGATION, TIMING... Xona Space Systems (s/o our own Jeff Crusey, who led their seed back in 2021) is moving the Pulsar satellite into rate production, as its first in-house flight units await liftoff on SpaceX’s Bandwagon-5 (NET 10/19) and four more enter production. "There is a point with every breakthrough technology where the challenge shifts from proving that it works to proving that you can build it again and again. Pulsar has reached that point," says Patrick Cadariu, who spent nearly a decade scaling mfg/supply chain at Waymo and recently joined the PNT-from-LEO startup (#047) as COO.
003 / ARRAY LABS … With great power comes great responsibility, and when your editor’s company books a win, he’s got to use the pulpit. Last week, Array Labs picked up a $4.47M award from the Army Research Laboratory (DEVCOM, for the acronym-fluent) to develop the distributed radar clusters that make 3D imaging, and other capabilities, work from orbit. Cue the meme…
004 / MUELLER'S MOVING CO... Impulse Space, backed by our friends at Cyrus Ventures and Overmatch Ventures, topped up its Series D with $308M, now at $808M on a reported $5.4B. The company has sold out its Caravan 2 and 3 rideshare missions for 2028, passed 600 in headcount, and it has 210+ open roles (apply & tell ‘em we sent you!) Impulse also notched a $28M USSF contract for two more Mira missions. S/o the whole team and Tom Mueller, founder/CEO, SpaceX Employee #1, and disciple of FBC — as he tells us himself!
005 / SHEP, HONORED... Bill Shepherd, the former Navy SEAL that Dan picked to be the first ISS commander (#018), is this year's Spirit of Discovery honoree at the Cradle of Aviation Museum's Air and Space Gala on Nov. 12 in Long Island. Well done, Shep!
🤠 Big s/o to dear friend of PA Jordan Blashek (catch our Violence of Action Q+A) and the whole Overmatch crew, who just dropped the first annual Austin Deep Tech 50. Eight funds cast the votes (Overmatch, Valor, 8VC, Banner, B Capital, Ambition, Ensemble and Trust Ventures), and to make the cut you had to be pre-IPO, building deep tech, and based in Austin or with a major presence there.
It reads like our own back catalog, with names like Base Power, Saronic, CesiumAstro, and many more that you’ll recognize. All told, the 50 have raised $10B+ in the last twelve months across energy, robotics, manufacturing, maritime, biotech, compute, and defense.
“The companies rebuilding and reinventing America’s power grid, defense industrial base, and robotics are doing it in Austin,” says Overmatch’s Aabid Razvi. Plenty of y’all are living proof. Congrats to all 50, and well-done to the team who put it together 🙌
With that, we’ll see y’all back here next week.
|