This essay is entirely non-defense related and serves as my long form “out of office” message before I depart for a month of moving and traveling. If this doesn’t interest you, feel free to stop here. But if you want to be moved by stories of childish foolishness, images from the Hudson River School art movement, and the importance of liminal spaces in life, then by all means keep reading.

The Encierro

The sun was rising in Pamplona-Iruña, and the roar of the crowd that had gathered for the Encierro was deafening.

Everyone, including my friend Preston and me, wore the iconic combination of red and white that characterized the participants of the annual running of the bulls, for which Pamplona is famous.

Police were moving through the crowd, pulling aside participants who were too drunk or tired to run safely (yes believe it or not there is some safety involved). The custom of the encierro is to spend the entire night drinking and partying before moving to run with the bulls at sunrise. (Steves, 2026) Despite being 20 years old and in our prime, Preston and I opted to survive the night by remaining as sober as possible, concentrating instead on the run ahead.

This was less a decision based on good judgment, and rather it was because after a long summer in Spain we were completely and unequivocally broke. With only a return ticket to our name, we relied on bread, water, and the kindness of strangers to survive the remaining week until our return.

The police moved past Preston and me, in search of offenders to remove from the crowd.

The sleeping man in question. Apparently I had the wherewithal to take a photo instead of providing actual help. Again, I was 20, so forgive me my transgressions.

Meanwhile, you could hear the bulls bumping against the gates we had foolishly placed ourselves against. We would only learn afterward that for a good run, you never start close to the gate, or at least not if you hope to make it into the stadium at the end. But we didn’t know that then.

Instead, we sat next to the gate and watched the Bravos — the large black bulls, the truly fearsome ones — mingle alongside the smaller, more docile white bulls that served as their guides. Beside us, a man who’d been standing since sunrise finally lost the fight with sleep and pitched forward onto the cobblestone, drawing the police away just before the gates opened.

Perhaps only a minute later, at 8:00 am, two rockets fired into the early morning sky, signaling the beginning of the race and the release of the bulls.

Whatever risk-taking ambitions had driven Preston and me to race that day were quickly replaced with pure and unadulterated fear. There are no words that I can write here that can fully convey how terrifying the combination of these bulls’ size and speed can be. These are massive animals, weighing in excess of 1,200 lbs, capable of running 25 miles per hour in a straight line, though the course’s sharp turns (the dogleg at Mercaderes especially) force them to slow just enough for the pack to survive it.

The full race is quick. It lasts somewhere between two and a half and three minutes.

For the most experienced of runners, the “perfect race” is one where, having the bull’s horns squarely behind you (”running on the horns”), you can feel the hot breath of the Bravos on your back for 20-30 seconds before expertly diving off into safety.

I would never claim to be an expert, but I certainly felt the bulls’ breath that day. Although if I actually lasted a full 30 seconds before diving off onto the fence, I would be shocked. Adrenaline has a funny way of searing some things into your memory but washing the rest away completely.

Preston and I have obviously lived to tell the story without being gored by a bull. We celebrated that morning with a churro con chocolate. Calories after near-death experiences don’t count.

We made grand plans that only 20-year-olds with no conception of the demands of the real world could make. The kind of plans that only look foolish once experience has taught you what is real. We’d come back to the north of Spain to hike the Camino de Santiago. After that, somewhere new every year.

Almost 14 years later, I found myself sitting in a Harvard classroom listening to Arthur Brooks talk about his own career and life.

Subscribe now

Running the Horns

He began with a picture of himself on a trail in the woods, and recounted for us how, after stepping down as the CEO of a think tank in Washington, D.C., he had taken advantage of that time and space to hike the Camino de Santiago.

Looking at that photo, I thought of a different picture. One I’d been carrying around for years without quite knowing what to do with it. Before coming to HBS, I lived in Washington, D.C., and one afternoon I stumbled into the National Gallery of Art and found myself standing in front of Thomas Cole’s The Voyage of Life, a set of four paintings tracking a single figure through childhood, youth, manhood, and old age. The set as a whole stopped me, but it was the picture of Youth that stayed with me. In it, a young man stands in a boat on a clear river, pointing toward a shimmering castle suspended in the clouds ahead of him.

Cole himself, writing about the painting, was under no illusion about what that castle actually was:

“The scenery of the picture — its clear stream, its lofty trees, its towering mountains, its unbounded distance, and transparent atmosphere — figure forth the romantic beauty of youthful imaginings, when the mind elevates the mean and common into the magnificent, before experience teaches what is the Real.”

I can practically hear the disdain in that last clause: before experience teaches what is the Real. Cole knew the castle wasn’t real. He painted it anyway, because the wanting of it is the whole point of being 20 and broke and sitting against a bull gate in Pamplona with no idea what you’re doing.

The third painting in the set, Manhood, is shows the same boat and man worn and destroyed. The same river turned to whitewater, the rudder gone, the figure gripping what’s left of the hull and praying. A decade flying helicopters and driving warships through the Strait of Hormuz gave me a fairly literal education in what that painting is describing. Fatherhood, in its own way, has given me another. Most days feel closer to Manhood than to Youth. But every morning I still try to wake up chasing the castle anyway. I’ve come to believe that the day you stop reaching for it is the day you start aging on purpose.

Check out the full collection of Thomas Cole here.

Brooks has a word for the water between those two paintings. Liminality. He takes it from the psychologists, and he means the time between things: between roles, between organizations, between career paths, and between one version of yourself and the next.

We have names for the chapters in our minds: midshipman, aviator, student, or whatever comes next, but nothing for the white space in between, so we treat that space like an accounting error. Something to be compressed. You resign on a Friday, you start on a Monday, and you tell yourself you were being efficient.

Brooks argues that you do the opposite. The gap is uncomfortable by design: you’ve set one identity down and haven’t yet picked the next one up, and there’s nothing in your hands.

The questions he poses for this stretch of river are almost insultingly simple. What do you keep? What do you carry forward but do differently? What do you put down for good? What do you learn that you’ve never done before?

He has a companion argument about what to actually do with the space once you’re standing in it, and the answer is not to rest. Openness to new experience (what psychologists sometimes call neophilia) functions as its own engine of interest. Having your curiosity provoked is pleasurable in itself, and the people who keep putting themselves in front of unfamiliar things trip that wire more often, and harder, than the people who don’t.

An adventure requires you to be bad at something in public: a new country, a new language, a new trade, a new set of people who haven’t read your résumé and wouldn’t be impressed by it if they had. For anyone who has spent a decade accumulating competence, that’s the actual price of admission, and it’s why most people in transition book a vacation instead of an adventure. A vacation is a rest from your identity. An adventure is a test of it.

To this day, the feeling of running on the metaphorical horns of the bull, the hot breath at your back, the couple of seconds before you have to decide whether to hold your ground or dive for the fence, is one that I rather enjoy, and one I try to keep seeking out.

The Road Goes Ever On

I spent ten years flying helicopters and driving warships through the South China Sea, the Middle East, and the Eastern Mediterranean. I’ve tracked submarines. I’ve run interdictions on Houthi weapons smuggling operations off the coast of a country most Americans couldn’t find on a map.

When I left the Navy for business school, I told myself I wanted a simpler, quieter life. In some ways I got one. What’s painfully clear two years later is that the off switch for that feeling is much harder to find than I expected. I work from sunrise to sundown on any number of things. I take on too much, and I love almost all of it. Neither my family nor my friends would testify otherwise.

But I think it is high time I took a detour for another adventure and put the work aside for a while.

So as for how I’m using this stretch of open water: after 14 years of delayed promises, I’ll finally take that trip to northern Spain and hike the Camino de Santiago. This time I won’t be a broke college student. I’ll bring slightly more resources, better judgment, and a fair amount of lived experience with me on the trail.

As I write this, I sit among moving boxes, packing for our return from Boston to Washington, D.C., where we’ll start the next chapter of our careers and lives. The chapter of graduate school and military service is closed. We’re turning the page.

Before the words of the next chapter start, though, I intend to make good use of this liminal space. With that, I’m putting on my out-of-office reply for the next month. Any writing I do will live on a notepad, not on Substack, and I know my readers will forgive the silence even if the algorithm won’t.

What the adventure holds I’ve already told you. What the next chapter holds, I’ll leave for another time. I’ll close instead with two lines from Tolkien (a clue perhaps?), borrowed and true enough on their own:

The Road goes ever on and on
Down from the door where it began.

Far ahead the road has gone. I mean to follow it, and see where it joins the larger way.

Share

Archimedes famously said, “Give me a place to stand, and a lever long enough, and I will move the world.”

In today’s defense industrial base, that place to stand is the foundational architecture of Large Language Models. The lever is Answer Engine Optimization (AEO).

For decades, shifting the narrative in national security required the massive gravitational pull (and lobbying budgets) of legacy primes. Men and women in power suits have long stormed Capitol Hill with their glossy one-pagers and gratuitous compliments, cornering underpaid, over-caffeinated 20-something legislative assistants in search of favorable bill text or a sliver of appropriations. Meanwhile, policy analysts, locked in windowless Pentagon SCIFs, relied on Google, closed databases, and trade publications to craft the well-honed memos that guided decision-makers.

But the game has fundamentally changed.

Today, those same staffers and analysts leverage AI to synthesize procurement research, draft policy memos, and summarize congressional hearings. The outputs of these models have a growing, outsized impact on how Washington understands the world. Consequently, optimizing how a company or policy position appears in these generated answers provides a sheer, undeniable mechanical advantage.

I will be the first to admit it: I lean heavily on Claude and other models as my default starting point for research. As much as I pride myself on asking the right questions and hunting down primary sources, it is impossible not to be nudged—even slightly—by a model’s initial output.

That realization led me down a rabbit hole. I wanted to understand exactly how defense disruptors show up in these responses.

What I found is that by optimizing how algorithms perceive and summarize capabilities, today’s startups don’t need brute mass to move the market.

They just need to apply pressure at the precise fulcrum of institutional sentiment.

The Rising Complexity of Defense Tech

What is AEO at its simplest? It is the deliberate optimization of how your company appears in Large Language Model (LLM) and AI chatbot responses. It is the successor to SEO (Search Engine Optimization), but with a radically different endgame. Where SEO fought for a click on a blue link, AEO fights for a direct citation in an AI’s synthesized answer.

If you are a long-time professional in national security and you never fully bought into SEO, that is not surprising. Until recently, SEO was largely unnecessary in aerospace and defense. There were essentially five major primes, and the tier-two suppliers had long-standing, locked-in relationships with them. You did not need to “search” for who built what; you already knew. In fact, it was practically a mark of professional prestige to know the intricate ins and outs of these companies without having to look them up.

But the confluence of the AI revolution and the explosion of defense tech startups has created an unprecedented nexus of complexity.

The number of defense tech startups emerging year over year is accelerating. Venture funding into defense tech companies continues to set records, and with major gaps in our readiness emerging (think munitions and low cost interceptors) alongside daily innovations on the Ukrainian battlefield, more founders and investors are charging toward this space every day.

Take the Golden Dome IDIQ as the perfect illustration. When the Pentagon announced the initial awards under the SHIELD contract for the Golden Dome missile defense program, they did not just select a handful of traditional primes. They awarded seats to 1,014 qualifying vendors out of more than 2,400 offers for a contract vehicle with a $151 billion ceiling. By January 2026, that pool had grown to 2,440 approved vendors out of 2,463 applicants.

If you are hypothetically the Chief Contracting Officer or a PAE for Golden Dome, how do you stay abreast of product capabilities, technical milestones, and corporate pivots from a list of over 1,000 qualified OEMs and suppliers? It would take an entire division of analysts just to track the roster.

Or perhaps just one enterprising analyst using GenAI.mil?

The Mechanics of AEO: The RAG Pipeline

This is where the shift in AI completely rewrites the procurement landscape. When a PAE asks their AI assistant, “Who are the most qualified vendors for Golden Dome’s space-based interceptors?”, the LLM draws on two distinct bodies of knowledge to formulate its answer:

• Pre-Training Data: This is the historical information baked directly into the model’s foundational weights during its initial training. It is essentially static and almost impossible for a company to quickly manipulate.

• Retrieval-Augmented Generation (RAG): Because the defense tech ecosystem moves too fast for static models, AI tools actively search the internet for real-time updates to augment their responses.

The RAG pipeline is where the battle of perception is won or lost. To answer the prompt, the model pulls real-time data from third-party sources it deems highly credible: articles from DefenseScoop, award announcements on DIU.mil, and reputable industry Substacks (cough cough IBA). AEO is about ensuring your company’s capabilities are consistently and favorably represented in those exact external citations, ensuring the LLM pulls your narrative to construct its answer.

The Buyer Side: How Algorithms Shape the Battlefield

If you’ve made it this far, you might be thinking: Okay, cool. But how does this apply to me?

Let’s look at the buyer side of the equation. At its core—though many won’t readily admit it—the smartest people in Washington are turning to Large Language Models to bring themselves up to speed. The individuals positioned to make highly consequential decisions about NDAA amendments, defense appropriations, or programmatic shifts within the Pentagon are all beginning their research in a chat window.

I will be the first to admit that I do most of my initial research within Claude, backed up by other models, before I turn to primary sources to verify the facts. I have a series of items on my diligence checklist that I run through using Cowork (e.g., company history, competitive landscape, reported financials, etc). Then I go pull sources directly for more color and to verify the authenticity of factual statements. But even before I dig into those third-party sources, I am largely biased by the model’s initial output. If I want a list of the most qualified counter-UAS companies, my first impression is formed by Claude’s response for better or worse.

I am hardly alone. Recent reporting on AI adoption inside Congress has been staggering; the overwhelming majority of House and Senate offices are now spending heavily on tools like ChatGPT Plus and Claude to summarize legislation, draft memos, and prepare hearing materials. CNBC reported this week that ChatGPT dominates congressional spend with 88% of all dollars going to AI tools. And for those who haven’t watched this iconic video of a Canadian politician reading his AI generated talking point, it is worth a watch.

Now, combine this AI adoption with the ongoing wave of acquisition reform. How the Department of Defense procures systems is changing. The use of Other Transaction Authorities (OTAs) and sole-source contracts for startups is gaining serious momentum. While the raw numbers of these alternative pathways are still relatively low compared to traditional FAR-based contracting, the trendline is steep.

What is significant about this shift is that fewer people are making more consequential decisions with taxpayer dollars, and they are moving at a faster pace. Generally, this is a massive win for defense innovation. But an unintended consequence is that PAEs and contracting officers are far more susceptible to being nudged toward a particular vendor if their foundational research begins with an LLM.

Before your lobbyist and business development team ever badge into the Pentagon or walk the halls of Congress, these decision-makers have already formed an opinion about your company.

Access used to be a function of paying exorbitant retainers to recently retired flag officers or former administration officials with deep Rolodexes in programmatic offices. I am not saying that traditional access is dead—a component of relationship-driven sales will remain for a long time to come.

But the long tail of your online presence and media footprint is constantly influencing these people’s impressions of your company. It happens day in and day out, while you sleep, while an overworked staffer stays up late researching the exact amendment you desperately need them to draft.

And yet, almost no one in defense tech is giving this the attention it deserves.

How to Actually Optimize for AEO

So, how do you actually optimize for LLM responses? The short answer is conventional SEO paired with earned media, authentic conversation, and a relentless representation of brand voice that trickles into third-party citations.

It is the founders, CEOs, and board members who are talking directly to the public through social media and giving exclusive access to third-party media companies who are showing up the strongest in LLM responses. Lockheed Martin may have the most formidable business development machine and the longest track record of success in the space, but they are far from the leaders in AEO. Rather companies with strong public personas like Palmer Luckey’s Anduril show up more in AI visibility (I have the data on this which I will release later so you have to trust me on this).

While this Substack’s North Star remains focused on the defense industry, the industrial base, and the people building incredible products for U.S. national security, it would be a mistake to ignore the larger trend. The benefits of AEO extend far beyond defense. Any company with a government affairs team, or anyone who benefits from a favorable regulatory environment, needs to think critically about how their policy positions, products, and services appear in AI responses.

The landscape is shifting fast. Think tanks used to be the undisputed centers of thought leadership; today, they are just one piece of a much larger ecosystem. If you want evidence of this playing out right before our eyes, look no further than the defense tech startups backed by Andreessen Horowitz (a16z) or other major venture capital firms. These VC arms now operate massive in-house media teams whose sole purpose is to elevate the brand, content, and beliefs of their portfolio companies. They have rightly tuned into how companies and products are discovered in a professionally online world.

If you are an engineer who believes you can just sit back, build a great product, and ignore marketing, you are going to be proven wrong.

Gone are the days when you could secure congressional favor with steak dinners. Gone are the days when think tanks were the sole purveyors of thought leadership. Gone are the days of hiring a retired Army Colonel purely for their legacy access to the program office they just left.

Consider the half-life of a traditional lobbying effort. A highly paid consultant drops a glossy one-pager on the desk of a legislative assistant in the Rayburn House Office Building. The moment that lobbyist walks out the door, the paper starts gathering dust. Its reach is strictly capped at whoever happens to glance at that specific desk on that specific day. When the K Street retainer runs out, the influence abruptly stops.

Media, on the other hand, is persistent. A well-crafted founder op-ed, a deeply researched Substack post, or a comprehensive FAQ page costs a fraction of traditional access, yet it lives in perpetuity. Naval Ravikant famously categorized code and media as the two forms of permissionless leverage. You don’t need anyone’s permission to publish, and once you do, that media works for you while you sleep. It gets indexed by search engines, and more importantly, it gets ingested by the RAG pipelines of AI models, continually shaping the narrative long after the money was spent to create it.

There is a new age being ushered in by AI and acquisition reform, and it relies on this central point: Media is the fulcrum and the lever that will fundamentally move this industry. My core argument is this: If your entire government affairs budget is going toward people and not toward media, you are playing catch-up. You are letting an algorithm tell your story before you even enter the room.

Opening Pandora’s Box

I would be remiss if I didn’t point out the ethical issues inherent to AEO. The power to nudge the masses and key decision-makers in the direction of your beliefs is profound.

AEO has the capacity for enormous good, but also enormous harm. Weaponized misinformation can be devastating. Bad information, once injected into the ecosystem, can produce hundreds of AI-generated derivative works that live on in perpetuity until forcefully erased by the truth. With any great leverage comes the responsibility to use it well.

This is not fundamentally different from traditional lobbying. Philip Morris using lobbyists to influence the congressional record is an example of influence used poorly; non-profits lobbying Congress to give a voice to marginalized communities is an example of it used well. AEO is no different. Just because it is driven by AI does not mean it should be eschewed, but we must tread carefully.

Where AI goes, what it is used for, and how we regulate it are all issues that still need to be settled. But anyone using these products today understands that we have opened Pandora’s box, and it will never be closed. AI and its synthesized responses will be a central part of life for anyone thinking deeply about policy, acquisitions, and national security.

I have already begun deep-dive research into this area, spending serious money on API tokens to benchmark AEO visibility for various defense companies. I will be publishing the first batch of that research (focusing specifically on the drone industry) later this week. It has already revealed some fascinating insights, but I wanted to keep that data separate to clearly outline the paradigm shift first.

If you are interested in how this affects your company or product lines, please reach out. I am passionate about helping the best of American engineering and business talent create incredible products for our national security, and a small but critical piece of that is ensuring you actually show up when the machine goes looking for you.

Share

To vertically integrate or not to vertically integrate.

That is the question.

It is asked in every pitch deck, every board meeting, and every deep-tech fundraisi…

Read more

Read more

Read more

According to NBC news:

Earlier in the day, a spokesperson for Iran’s Foreign Ministry, Esmail Baghaei, told re…

Read more

Four Republicans broke ranks to join nearly all Democrats on a House-passed resolution directing the President to pull U.S. forces out of the conflict with Iran.

This is the rare congressional check on a sitting wartime president, and the G…

Read more

With peace breaking out in the Middle East (allegedly), and my own writing for The Merge continuing (check out last week’s deep dive into Impulse Space), I wanted to bring you something new.

Each day, I analyze the three biggest contract announcements, acquisitions, and aerospace & defense headlines th…

Read more

Cyber is an area IBA…

Read more

For those who regularly tune in for defense industry analysis, this one is different. For those who are new here, don't worry, you didn't subscribe to the wrong newsletter. I started this Substack primarily to sharpen my writing and thinking, and after graduating from Harvard Business School this past week, I wanted to take stock of what these two years…

Read more

Astranis announced its Series E on May 6, 2026: $300 million in equity co-led by Snowpoint Ventures and Franklin Templeton, plus up to $155 million in delayed-draw debt from Trinity Capital. That is roughly $455 million in new capital, and it pushes Astranis past $1.2 billion raised to date.

The round valued the company at about $2.8 billion post-money, a 2.5x step-up from its 2024 raise.

What drew me to write about Astranis, other than the fundraising news was the mission. The company delivers low-cost internet to remote communities using small geostationary satellites rather than a sprawling low-Earth-orbit constellation, and it gives the US government a cheap, dedicated option for military communications. One satellite design, two markets.

I wrote about True Anomaly last week. Astranis is a different kind of company. True Anomaly builds for the Pentagon. Astranis is genuinely dual-use, with a commercial business and a government business that each stand on their own.

This week we are digging into the mission, the technology, the business model and what we can glean from public market comparisons.

This one is lengthy, but packed with analysis that I promise is worth reading. Buckle up and let’s get into it.

The Small GEO Bet

At the simplest level: Astranis builds small dedicated communications satellites for geostationary orbit (GEO).

Most companies in the satellite world today are racing to put thousands of small satellites in low-Earth orbit (LEO): Starlink, Kuiper, OneWeb, AST SpaceMobile.

Astranis is doing the opposite: one satellite, parked 22,000 miles up, serving one customer over one region.

Each Astranis satellite is roughly the size of a mini-fridge (400 kg), costs in the $15–30M range, and gets built in months. Compare that to a traditional GEO comms satellite from Boeing, Lockheed, or Airbus, which is the size of a school bus (6,000+ kg), costs $300–500M, and takes 5+ years to deliver. That’s the core engineering bet: they shrunk GEO.

Source: Astranis via Techcrunch

How they’re differentiated

• Against legacy GEO primes (Boeing, Lockheed, Northrop, Airbus, Thales Alenia): In addition to being a fraction of the size of a traditional GEO satellite, Astranis is roughly 10x cheaper and 10x faster. The primes are still selling bespoke flagship satellites; Astranis is selling something closer to a product. They’ve vertically integrated the entire stack in San Francisco, running the same playbook as SpaceX and Anduril..

• Against LEO constellations (Starlink, Kuiper, AST SpaceMobile, OneWeb) — Starlink, Kuiper, AST SpaceMobile, and OneWeb all sell consumer and enterprise broadband from low-Earth orbit. The capacity is shared. The geopolitics are messy. Your bandwidth depends on a constellation owned by someone else. Astranis sells dedicated capacity over a single country. The Chunghwa Telecom deal in Taiwan and the MB Group deal in Oman are sovereignty plays as much as they are broadband deals.

• Against legacy GEO operators (Intelsat, SES, Viasat, Eutelsat) — These guys lease shared transponder capacity on satellites they operate themselves. Astranis sells you the entire satellite. CEO John Gedmark cited “increased demand from governments and enterprises seeking dedicated communications infrastructure and moving away from shared geostationary communications satellites.” This is for Sovereign nations and large enterprises that want to own an asset on orbit rather than rent capacity from someone else.

The bottom line is that in Small GEO, there is Astranis and no one else.

Terran Orbital, Millennium Space, York Space, Apex. Every other notable small satellite builder is focused on low-Earth orbit.

Astranis is essentially alone in the small GEO category, with 5 satellites on orbit, a commercial backlog above $1 billion, and 500 employees.

Partner With Industrial Base Alpha

What Astranis Gives Up to Stay in GEO

Surely everyone else who is building in LEO for telecommunications is not wrong. There must be some trade offs to operating in GEO. Well it turns out that there are quite a few notable commercial disadvantages.:

• Latency is a deal-killer for half the use cases. A GEO round-trip signal takes 480 to 600 milliseconds versus 25 to 40 ms for LEO. That gap rules out real-time video conferencing, online gaming, algorithmic trading, and any tactical application where milliseconds drive targeting decisions.

• Per-satellite capacity is fixed and modest. An Astranis MicroGEO delivers roughly 7 to 10 Gbps of throughput. A Starlink V2 Mini delivers 80 to 100 Gbps, and a LEO constellation aggregates capacity across thousands of birds with frequency reuse on every spot beam.

• The technology is frozen at launch for 15 years. A 2026 Astranis satellite will still be operating in 2040 with 2025-vintage silicon, antennas, and power systems. Software-defined payloads help with frequency reconfiguration but do nothing about aging hardware.

• Launch and orbital slots cost more. GEO transfer orbit launch runs roughly $15,000 to $25,000 per kg versus $2,000 to $3,000 per kg for LEO. New GEO entrants also have to negotiate ITU orbital slot rights against incumbents like Intelsat and SES, which is a coordination tax LEO does not pay.

Militarily, there are also a couple of vulnerabilities:

• Polar coverage does not exist. GEO satellites sit over the equator and become unusable above roughly 75 degrees latitude. Arctic operations, far northern Canada, northern Russia, and Antarctica are permanent blind spots for Astranis.

• One satellite is one target. A LEO constellation with 5,000-plus satellites is functionally impossible to kill at scale. A single GEO bird is a defined target, and the Pentagon’s proliferated LEO push exists specifically to address that vulnerability.

• Ground terminals are bigger and less mobile. GEO terminals typically need a parabolic dish 60 to 90 cm across to close the link budget against the 22,000-mile signal path. A Starlink phased-array terminal is the size of a placemat: the entire reason it works for vehicles, ships, and tactical units.

So why is this still a real business? Three reasons.

Read more

Graphics like this make me think that marketing for space companies has to be the easiest job in the world. Is it possible to make bad marketing material of space?

BLUF: True Anomaly's $650 million April 2026 Series D is the third-largest defense-tech round on record. By month 52 from founding, the company has raised 4.7x what BlackSky raised at the same…

Read more

Industrial Base Alpha now offers institutional defense market research and intelligence services for investors, acquirers, and corporate strategy teams. If you need public market context, sector comps, or procurement-informed analysis to support investment and acquisition decisions, we'd like to hear from you. Fill out this short Google form to start a …

Read more

Huntington Ingalls Industries said Tuesday that first-quarter revenue rose 13.4% to $3.10 billion and diluted earnings per share were $3.79.

Both numbers beat Wall Street estimates. Shares fell about 5%, to roughly $345.

The Newport News, Va.-based company is the largest pure-play military shipbuilder in the United States. It builds the Navy’s nuclear aircraft carriers at Newport News Shipbuilding and amphibious and surface combatants at its Ingalls yard in Pascagoula, Miss. It also contributes to the Virginia-class and Columbia-class submarine programs alongside General Dynamics’ Electric Boat. A third unit, Mission Technologies, is a defense-services business.

The first-quarter results were the company’s strongest top-line print in five years. They were also accompanied by a 90-basis-point drop in operating margin, to 5.0% from 5.9% a year earlier, and a negative $461 million in free cash flow.

So if HII beat expectations then why did the stock drop like a rock?

Investors focused on three numbers.

The first was operating income, which fell $6 million even as revenue rose $365 million. Cost of product sales grew 20% against 17% product revenue growth. Huntington Ingalls said the margin compression came from lower performance on aircraft-carrier construction at Newport News and from lower equity income at Mission Technologies’ nuclear and environmental joint ventures. Submarine revenue at Newport News rose 21%, aircraft-carrier revenue rose 16%, and surface-combatant revenue at Ingalls rose 26%.

The second was the composition of the earnings beat. Net income of $149 million was flat with the prior-year quarter despite the 13% revenue gain. A $5 million increase in non-operating pension benefit, driven by 2025 pension-asset returns, and a $6 million decline in interest expense, driven by debt paydown, kept earnings per share level. Strip out those below-the-line items and operating earnings declined.

The third was a line on the balance sheet called net contract assets. It represents work the company has performed but for which the U.S. government has not yet billed or paid. The figure rose by $629 million in three months, to $1.17 billion. Huntington Ingalls said the increase reflected the timing of billings on certain Navy contracts.

Partner With Industrial Base Alpha

Read more

SpaceX filed confidentially for its IPO last month at a reported valuation of $1.75 trillion. The filing dominated news cycles for weeks. Analysts called it the most consequential capital markets event in a generation.

There is not much to learn from that. Elon Musk’s companies operate in a category of one. The scale, the mythology, the gravitational pull on investor attention, and none of it translates into a playbook anyone else can follow.

HawkEye 360’s April 10 IPO filing is a different story. The below statement from Hawkeye 360 CEO John Serafini on the Mission Matters Podcast (produced by Shield Capital) sums up the importance of this moment:

“What we’re really missing now…is the exit opportunities, particularly in space. That’s always what’s held back commercial space is a lack of exits,”

Founded in 2015 in Herndon, Virginia, HawkEye 360 is a radio-frequency signals intelligence company that took 11 years to reach the public markets.

It did not SPAC. It did not raise at a valuation it could not grow into. It built a satellite constellation, won government contracts in sequence, made two acquisitions to complete its technology stack, and turned a profit before it filed.

For founders and investors in defense and space, the HawkEye filing is a roadmap (full roadmap at the bottom).

Note: The bulk of the financials here come from the 200+ S-1, which I highly recommend reading here at the SEC website.

Who is HawkEye 360?

HawkEye 360 was not built in a garage.

The core technology originated in a Virginia Tech research lab, where engineers developed the foundational algorithms for detecting and geolocating radio frequency emissions from space.

Allied Minds, a Boston-based firm that specializes in commercializing university and federal lab technology, identified the opportunity and spun the company out in 2015 (the Allied Minds minority stake was later sold off in 2019 so no connection remains today).

That origin matters for anyone working on dual-use technology inside an academic or national lab today: the path from bench to battlefield does not require a Silicon Valley pedigree.

The company flies triads of satellites, referred to as clusters, across multiple orbits to detect and geolocate radio frequency emissions from communications systems, navigation devices, and radar sources.

Proprietary signal processing and AI-powered analytics convert that raw collection into finished intelligence products.

As of early April 2026, HawkEye has 14 clusters on orbit. The capability covers a broad operational range: maritime domain awareness, GPS jamming and spoofing detection, handheld device communications mapping, and radar characterization.

RF intelligence layers onto data generated by electro-optical and synthetic aperture radar sensors, filling gaps those systems cannot address.

Hawkeye’s Vertically Integrated Technology Stack (Image from S-1 Filing)

HawkEye spent its first decade in the long middle of the defense acquisition cycle.

The company began winning exploratory contracts with the NRO and NGA in 2019, then expanded to AFRL, INDOPACOM, and allied defense ministries across three continents.

For several years, the contracts were real but the dollar amounts were not yet large: prototype agreements and proof-of-concept work that validated the technology without converting into sustained program revenue.

That period is where most defense-tech companies stall. HawkEye did not.

The cap table was stacked with investors who understood government procurement timelines, the C-suite kept the burn disciplined, and the company kept winning the next contract.

Then, in 2025, the exploratory agreements gave way to funded programs. Revenue grew 74 percent year over year.

Partner With Industrial Base Alpha

Capability Consolidation via Acquisition

HawkEye’s path to vertical integration began not with a purpose-built acquisition target, but with an orphaned asset inside a much larger company.

In 2023, Maxar Technologies was in the process of being taken private by Advent International.

Its RF intelligence unit, originally built through Maxar’s own acquisition of Aurora Insight, had become a non-core asset in the transaction.

HawkEye moved quickly. The deal added two satellites to HawkEye’s constellation and, more importantly, brought Ka-band monitoring capability into the platform for the first time.

Ka-band covers the frequency range used by modern radar systems, high-throughput satellite communications, civilian 5G infrastructure, and weather systems.

A single transaction extended HawkEye’s physical coverage and meaningfully expanded the intelligence picture it could deliver to customers (S-1, p. 12).

The second acquisition was more consequential. In December 2025, HawkEye acquired Innovative Signal Analysis, a signals processing firm founded in 1997 with nearly three decades of work inside U.S. defense and intelligence community programs.

The strategic value of ISA was not its hardware. It was the algorithms: signal characterization models built and validated over 28 years against the specific data environments, classification requirements, and processing architectures that HawkEye’s government customers operate within.

Buying ISA gave HawkEye the ability to operate on both sides of the classification wall, converting a strong unclassified data business into one with classified processing capability and the institutional relationships to match.

The acquisition was financed with debt, a decision that would factor directly into the timing of the IPO filing four months later (S-1, p. 18).

Refer a friend

Hawkeye’s 11 Year Fundraising Journey

The SPAC boom of 2020 and 2021 produced a specific kind of defense-tech company: one that raised hundreds of millions from retail investors on the strength of a slide deck, a vision, and a total addressable market estimated at a number large enough to make the valuation feel reasonable.

Momentus, AST SpaceMobile, Spire Global, BlackSky Technology, and a half-dozen others went public via SPAC during that window all in search of liquidity during a time that the private markets had all but frozen over. Most were pre-revenue or barely past it. Several had not yet launched a functional satellite. The market assigned them enterprise values in the hundreds of millions anyway.

The correction was swift. By 2022, many of these companies had lost more than 80 percent of their post-SPAC share price. The underlying problem was not fraud. It was sequencing. Capital arrived before the technology, the contracts, and the customer relationships were ready to support it.

HawkEye 360 was raising money during the same period, but instead of taking the easy liquidity that a SPAC would provide, they chose the longer and harder route to their traditional S-1 filing on April 10th, 2026.

There are many lessons that a newer space company or investor can learn, but the main one here is that each round tracked a specific operational unlock.

The Series A in 2016 funded the Pathfinder cluster. The $70 million Series B in 2019 followed the company’s first product launch, RFGeo, and its first NRO contract. The $55 million Series C in 2021 funded three additional satellite clusters and arrived after the company had demonstrated that its data had real government buyers, not just letters of interest. The $145 million Series D closed the same year, as the contract pipeline began converting from exploratory agreements to funded programs.

This is a fundraising posture worth naming (and one not always followed despite its proven success): raise to the next milestone, not to the terminal vision. Build the capability stack brick by brick, and don’t allow an ego-driven valuation hunt to drive a nail in the coffin.

HawkEye did not raise a $500 million Series B on the argument that the RF intelligence market would reach $34 billion by 2030. It raised $25 million in 2016 to put satellites in orbit. Then it raised more when those satellites produced data that customers paid for.

Finally, its worth noting how many investors and individual rounds it takes to get here: HawkEye had 42 investors across 8 rounds by the time it filed its S-1. The final round before the IPO filing was led by NightDragon, a cybersecurity and defense-focused fund that first invested in HawkEye in 2021, and Center15 Capital. Repeat investors across multiple rounds are visible throughout the cap table.

This is the sort of financial coalition founders should be building along the way.

Hawkeye’s Diversified Revenue Streams

HawkEye 360 closed 2025 with $117.7 million in revenue. 39% of it came from outside the United States.

The S-1 breaks the customer base into three buckets: U.S. government entities at 61%, Japan at 16%, and all other international customers at 23%. For a company that spent its first decade building trust inside the U.S. intelligence community, that last number is a structural achievement.

Image is from our own financial notebook created for this analysis.

The U.S. anchor is the National Reconnaissance Office, HawkEye’s largest federal customer, joined by the NGA and Space Force. In December 2025, the NRO renewed its relationship through a new contract to expand RF data and analytics for tactical use. The U.S. Navy renewed its Indo-Pacific maritime awareness contract the same month. Two of the most demanding customers in the U.S. government renewed in the same thirty-day window.

The 39% from outside the United States reflects a deliberate choice, not opportunistic contract wins. Many allied governments lack sovereign signals intelligence capacity. They cannot build the satellites, train the analysts, or develop the processing algorithms. HawkEye’s unclassified data model was designed from the start to be shareable across coalition partners in ways that classified government systems cannot.

In May 2025, the Defense Security Cooperation Agency approved a Foreign Military Sale to India for HawkEye 360 technology to enhance maritime domain awareness in the Indo-Pacific.

Japan at 16% of revenue is another safe and stable diversification of revenue streams. The sovereign-competitor risk is minimal. Japan’s defense posture, shaped by Article 9 and decades of structured reliance on U.S. intelligence partnerships, makes an independent commercial SIGINT program a remote scenario.

We ran a quick comparables analysis at IBA and here are some of major takeaways about what Hawkeye 360 should be worth.

Read more

BLUF: Private equity has found its role in the defense industrial base, and it is not replacing venture capital in defense technology. They are doing something more specific: taking companies that VC already de-risked, consolidating them into platforms with recurring government revenue, and returning them to the market at valuations that strategic acquirers and public investors will pay. The Arlington Capital/BlueHalo exit at $4.1 billion and the Madison Dearborn/Aevex IPO at a $3 billion opening market cap are the first clear proof points of that model working at scale. Venture capitalists investing in drone and autonomous systems companies today should treat these two exits as the most important data available on where their investments are ultimately headed.

On April 17, 2026, Aevex Corp. opened on the New York Stock Exchange at $25 per share, roughly 25% above its IPO price of $20.

One day earlier, Arcline Investment Management-backed Arxis debuted on Nasdaq, rising more than 38% on its first day and opening with a market value of more than $11 billion.

Two PE-backed defense companies. Two days. Fifteen billion dollars in combined market value.

I have been searching for the best way to write this article for a couple of months now.

I credit Jim Corridore at Pitchbook for delivering the spark for this deep dive:

“Pitchbook Data shows that 2025 had $5.6 billion in PE drone deals (77 transactions), and 2026 has already had $2.9 billion (25 deals).

Source: Please follow the incredible Defense PE expert Jim Corridore at Pitchbook!

PE is such a large part of our economy (about 14% of GDP) and one fraught with many public perception issues.

Really, this started as a question to myself more than anyone else: Does PE add value to this part of the economy (or any part of it)?

The result is the story below.

The Week That Made the Case

On April 17, Aevex Corp. opened on the New York Stock Exchange at $25 per share, roughly 25% above its IPO price of $20. The company had raised $320 million in the offering. Madison Dearborn Partners, the Chicago-based private equity firm that acquired Aevex in 2020, retained approximately 79% voting control.

One day earlier, Arcline Investment Management-backed Arxis debuted on Nasdaq, rising more than 38% on its first day of trading and opening with a market value of more than $11 billion.

Two PE-backed defense companies. Two days. Seventeen billion dollars in combined market value.

This week was the product of a specific investment thesis that a small number of PE firms have been executing quietly since 2019: find mature, revenue-generating defense technology companies, consolidate them into mission-area platforms, and hold them long enough for the Pentagon’s procurement priorities to catch up with what the platforms can do.

Quick Note: Arxis is primarily an aerospace components manufacturer (bearings, seals, electronic components) and not a drone platform like BlueHalo or Aevex. It’s a PE compounding story in defense industrials broadly, not the drone thesis specifically.

Partner With Industrial Base Alpha

What Private Equity Actually Does in Defense

Most observers of the defense industrial base understand two categories of capital. At one end are venture capital firms, which fund early-stage companies building new technology. At the other end are the primes, the large publicly traded defense contractors that manufacture weapons systems at scale under long-term government contracts.

Between those two ends is a gap.

VC funds typically hold positions for seven to ten years and need exits.

Defense primes move slowly, governed by procurement cycles and congressional authorization.

Neither institution is well-suited to the middle stage of a defense technology company’s life: post-product, post-initial-contract, but not yet large enough or diversified enough to attract a prime or go public on its own.

Private equity fills that gap. PE firms do not take technology risk. They take execution risk. They buy companies where the technology is proven and the government customer relationship is established, then invest in growth: additional acquisitions, headcount, infrastructure, and business development. They hold for five to seven years. They exit when the platform is large enough for a strategic buyer or public market.

What changed is the size of the market.

Refer a friend

Building BlueHalo: 18 Acquisitions in Five Years

In October 2019, Arlington Capital Partners, a Washington D.C.-area firm with a long history in government-regulated industries, acquired Aegis Technologies Group, a Huntsville, Alabama-based defense engineering company. Arlington did not announce a grand strategy. It announced a single acquisition.

Read more

Defense Catalyst Co-Founder & CEO Chase Contreras mountain bikes, surfs, has completed multiple triathlons, and a habit of playing chess while watching YouTube videos about geopolitical strategy.

He is not obviously a man destined for source selections and contract negotiations.

And yet, straight out of the Air Force Academy, he chose to become a contrac…

Read more

Read more

Read more