Where the Money Lands: Defense, Deals, and the Industrial Space Age

Zach Martin

Welcome to SPACE CAPITAL on KMKT IR Hub Radio. I'm Zach Martin, and this is the show about the space economy and the money behind it: launch, orbital infrastructure, spectrum, and the capital that makes all of it go. Real assets, real ideas, a brighter tomorrow. One piece of housekeeping before we start. Everything today comes from publicly reported developments and from Space Capital's published research and Summit conversations. It is not investment advice, and when I'm giving you my own interpretation instead of a reported fact, I'll tell you which is which.

Zach Martin

Here's the setup. If you've been with us for the last couple of episodes, The Allocator's Blueprint and the one on SpaceX, AI compute, and the new rules of space capital, you already carry the scoreboard in your head. The record first half. The SpaceX listing. The idea that in Launch+ the plus is the business. I'm not going to walk you back through any of that. Today is about what happened after the confetti. Because a funny thing happens when a market gets its first giant public currency and a few trillion dollars of attention. The conversation stops being about whether the money is coming. It becomes a conversation about where the money lands, and what that forces everybody downstream to do differently.

Zach Martin

So here's how I want to organize the hour. Space Capital sorts the whole economy into three technology layers. Infrastructure is the physical stuff, the assets you place in orbit or on the surface of another world. Distribution is the connectivity, the processing, and the networks that move and extract value from that hardware. Applications is where an end customer in a vertical market actually pays for something. Space Capital's curated dataset tracks about $488 billion invested across more than 3,000 companies since 2009. What I find most interesting about that pile is how rarely capital stays put. It migrates toward wherever the scarce asset is. For a while that looked like launch. Then it looked like spectrum. Then it looked like compute.

Zach Martin

My argument today is that between late summer and the fall of 2026, four stories that look unrelated on the surface are rewiring that migration all at once. The first is defense, and a doctrine shift from proliferation to maneuver, laid out at the Space Capital Summit by the people designing Golden Dome and the people building the vehicles. The second is a satellite company with thirty employees and no factory, which is challenging an assumption the whole industry copied from SpaceX. The third is the Space Force deciding to buy positioning and timing from commercial constellations rather than finish building its own. And the fourth is wheels on the Moon, a stretch of surface logistics that decides whether any other lunar business has anywhere to plug in. Running underneath all four is a reality check on orbital data centers, because every one of these stories bumps into the same question: what does it actually cost to get mass to orbit, and how often can you do it?

Zach Martin

Before we go in, I want to set down one piece of connective tissue, because every one of those stories is financed differently than it would have been a year ago. On March 11, Space Capital announced its fourth fund, crossing one billion dollars in assets under management. Founder and CEO Chad Anderson said the space economy has moved out of what he called its skepticism phase and into its industrial phase, and he noted that only 17 percent of venture firms ever reach a Fund IV. On May 6, the firm formed an intelligence partnership with Nasdaq, led on the Nasdaq side by chief revenue officer Jeff Thomas, to host the quarterly Space IQ briefings from the Nasdaq MarketSite in Times Square. And the 2026 Summit itself was held at that MarketSite, two weeks after SpaceX debuted on the exchange. A Lunar Outpost lunar terrain vehicle was parked out in Times Square. I'll come back to that vehicle, because it turns out to be a surprisingly good symbol for where the capital is going.

Zach Martin

Now the currency itself. Look at what the big strategic buyers did this year. Rocket Lab announced an $8 billion acquisition of Iridium, at $54 a share in cash and stock, expected to close around the middle of 2027. Space Capital's read on why that matters is spectrum and services, not sentiment. Spectrum is the one asset nobody can manufacture more of, and Iridium's L band licenses are globally harmonized, meaning they work everywhere on Earth instead of country by country. Add roughly 2.5 million subscribers and decades of government and safety of life contracts, and you have an installed base in direct to device, positioning, and aviation. In the same quarter, Amazon paid $11.6 billion for Globalstar on what Space Capital describes as the same logic. And SpaceX, which had made four acquisitions in its entire history before this year, none larger than $525 million, closed a $60 billion all stock deal for Cursor days after its debut.

Zach Martin

Here is my interpretation, flagged as such. When the biggest names in the sector can pay for things with their own stock, the rules of who gets to buy what change. A launch company that reaches orbit can now buy the thing that lives in orbit. That's the Launch+ logic playing out in deal form, and I'll leave the definition where we left it earlier. What I want you to notice is the other side of that coin. A deal paid partly in stock is a bet on the acquirer's stock. Rocket Lab's Iridium deal doesn't close for the better part of a year, and a lot can happen to an equity price in that time. Space Capital itself lists, among the press coverage of its quarterly report, a MarketWatch piece asking why the SpaceX stock is now cratering. I don't have the figures in front of me and I won't characterize the size of any move. The point is structural. Public currency is a gift when the market is generous, and a constraint when it isn't. Keep that in your back pocket, because it comes back at the end.

Zach Martin

So with that backdrop, let's go to the layer where money is arriving with the clearest mandate attached. It's a story about shields and spears, and about why the most interesting word in space defense right now is not interceptor. It's maneuver.

Zach Martin

Let me open with a line from Chad Anderson, the Space Capital CEO, as he started a session at the Summit. He told the audience, and I'm quoting, "A lot of how that's being done can't be discussed in a room like this. So we're not here for the secrets. We are here for the direction." That's the right frame for everything in this chapter. We don't know the classified details, and nobody on that stage was going to tell us. But the direction is public, and the direction is a big deal. Space Capital published the write up on August 25 under the title Shield and Spear: The Age of Maneuver.

Zach Martin

Two people sat across from Anderson. Marcia Holmes is Deputy Director of Golden Dome for America. In Space Capital's description she's the architect, helping define what the largest defensive space architecture the country has ever attempted is made of and who builds it. Eric Romo is President and COO of Impulse Space, and he's the builder, because Impulse makes vehicles that move. A disclosure: Impulse is a Space Capital portfolio company. So treat Romo's comments the way you'd treat any founder's comments about his own market. The thread running through the whole conversation was that in orbit, the shield and the spear increasingly run on the same hardware.

Zach Martin

Start with the threat, as Holmes described it, and these are her characterizations. She said Russia has fielded hypersonic and Avangard systems and has used them on the battlefield in Ukraine, and that China is, in her words, projected to field over 1,000 intercontinental ballistic missiles, doubling its arsenal. Cruise missiles, hypersonics, and ballistic missiles can now reach the United States directly. She said, "Our homeland is no longer a sanctuary." The problem she kept returning to is a sensing problem. Radars and interceptors rooted to the ground can't see the newest threats early enough. So she said you have to go to space to see them from launch all the way through their flight trajectories, and to move interceptors into position before a missile deploys its warheads or its decoys. Her phrase was that space based assets are the only thing that's going to enable that.

Zach Martin

Why is this possible now, when it wasn't before? Romo's answer began where every space conversation begins, with launch. He said launch costs and access to space are dramatically different than they've ever been. The last time the country seriously talked about intercepting missiles from space was the 1980s, when reaching orbit cost a fortune. A constellation of a thousand spacecraft in low Earth orbit would have been laughable a decade ago. Then Starlink and others made it routine. And that maturity trickles down to the people building the hardware. Romo said of the manufacturing base, quote, "I'm not doing this for the first time. I'm doing it for the second or third time and scaling up." That's a capital point hiding inside an engineering comment. Second and third generation manufacturers are cheaper to finance than first generation ones, because the risk that remains is production risk, not invention risk.

Zach Martin

Now the heart of it, the doctrinal shift. The Pentagon's strategy for low Earth orbit has been proliferation. Put up so many satellites that an adversary can't possibly take them all out. The logic is simple, and it works. If you lose a few of a thousand, the network still functions. But Romo pointed out that proliferation has a geometry problem. In his words, medium Earth orbit and geostationary orbit are, quote, "literally, a million times bigger than LEO." Blanketing orbits that vast with enough satellites to cover every point at every moment is, he said, "completely untenable." And here's the catch. Most US investment has historically gone to exactly those high energy orbits. GPS lives in medium Earth orbit. Missile warning and secure communications live in geostationary orbit. Those are a handful of extremely valuable assets, and they are now under pressure for the first time. He pointed to recent reports of three Russian craft maneuvering around a US imaging satellite over Ukraine as a sign that everything up there has to be able to move.

Zach Martin

Let me give you a football picture, and this is my analogy, not theirs. Proliferation is depth. You can lose a lineman and put in another one, because you've got a roster full of them. That's low Earth orbit. But the satellites that watch for missile launches or keep GPS running are more like a franchise quarterback. You can't have a hundred of him. You have one, maybe two. And if you can't add bodies to protect him, the answer is that he has to be able to move in the pocket, step up, slide out of the rush. That's maneuver. It's a different kind of protection, and it costs a different kind of money. It costs propellant, propulsion, autonomy, and the ability to reposition on short notice.

Zach Martin

Which is exactly what Impulse says it's flying. Its Victus Surgo and Victus Salo vehicles are both Space Domain Awareness missions, both sitting in the clean room now, and both built for fast, high energy repositioning. One is bound for a geostationary transfer orbit, and one for low Earth orbit. And in December, working with Starfish Space, Impulse completed a fully autonomous rendezvous in orbit. One vehicle approached another using a single camera, with no joystick involved. Romo drew a distinction that I think is the most important sentence in the whole write up. He said, quote, "The ability to reach out and touch an asset when it wants to be touched is one thing. When it does not want to be touched is another." And he was candid about the difficulty. Closing velocities run to thousands of meters per second, and hitting a target is not so much a bullet with a bullet as, in his words, maybe 100 times harder than that.

Zach Martin

Here's what I take from that for the Infrastructure layer, and again this is interpretation. For the last decade the capital story in satellites was about making satellites cheap and numerous. The maneuver doctrine adds a second premium, on vehicles that can change orbit quickly, which means propulsion and in space mobility become scarce inputs the way launch slots once were. That's why a company like Impulse, which builds the vehicles that move other things around, sits in an interesting place. It isn't selling a sensor and it isn't selling a bus. It's selling the ability to be somewhere else, quickly. In a defense context that's worth a lot. In a commercial context, the same capability is a way to reposition satellites and deliver them to the right orbit. That's the dual use point Romo kept returning to.

Zach Martin

Now the economics, because that's where defense stories usually either become investable or don't. A recent Congressional Budget Office estimate pegged space based interceptors at roughly $22 million a shot. Romo noted that in historical context that isn't outrageous, since ground based interceptors like SM three and the GBI run from $30 million to nearly $100 million apiece. But multiply $22 million by the thousands of interceptors a real architecture would need, and the number gets very large, very fast. Holmes was direct about the constraint. General Guetlein, who leads the program, has said publicly that if space based interceptors can't be bought affordably and at scale, they will not be part of Golden Dome. Read that again. A program with a trillion dollar defense budget in its orbit has told the industry that cost per unit is a pass or fail test.

Zach Martin

Her fix is procurement itself. Because this is net new, she said, the program can start from a clean sheet rather than fifty years of cost plus habit. Quote, "We're giving you prize based competitions." That means rewarding demonstrated performance instead of paperwork, and dealing directly with builders to compress timelines. Romo, understandably, was a fan. He said, "You set an ambitious goal, and if you get it, you get paid a bunch. If you don't, sorry." Now think about what that does to capital allocation. In a cost plus world, the government funds development and the contractor carries almost no risk. In a prize world, the company funds the attempt and gets paid when it succeeds. Interpretation time: that moves the financing of demonstration from the taxpayer to private investors, which is exactly why venture and growth capital are suddenly relevant to a defense program in a way they historically weren't. The reward is large if you win. The loss is yours if you don't.

Zach Martin

The budget numbers around this are real, but they need care. Space Capital's research notes that US defense allocations for fiscal 2026 stand at roughly one trillion dollars, with $1.5 trillion proposed for fiscal 2027, explicitly prioritizing space, cyber, and AI. Proposed is the operative word. Congress has not finalized appropriations for that fiscal 2027 figure, and the endurance of prize based acquisition depends on sustained backing from the administration. Romo said as much himself. He cautioned that the early demonstrations aren't what pays off, long term production is, and confidence in it, quote, "will ebb and flow with news cycles and OMB reports." Holmes said Golden Dome is meant to field real capability by 2028, and both agreed the signal to watch is not press releases. It's contract awards, test campaigns, and the steady expansion of the space budgets that quietly underwrite it.

Zach Martin

I found the passage on speed and decision time the most sobering. Holmes said strategic assets in space actually, quote, "buy back time." Romo made the stakes concrete. Intercepting a missile in its boost phase means detecting a launch, relaying it, getting a human decision, and reaching the target, all inside a window measured in seconds, against something that may be accelerating through the atmosphere behind a wall of plasma. Persistent overhead coverage is the only way to compress that chain, which is itself the argument for a large, maneuverable fleet. Hypersonics are harder still, he added, because they stay in the atmosphere and can bank and dive away from an interceptor. Holmes summed it up simply: these are hard problems, very hard problems. And she noted the goal is increasingly not purely kinetic. It's about giving decision makers more time and better options, with AI surfacing recommendations faster.

Zach Martin

So where does an investor or a founder look? Anderson asked the question the firm came for: between the architecture Holmes is defining and the vehicles Romo is building, where is the white space? The answer was a layered defense running from space to land to sea, and a long list of pieces still to be built. Low cost kinetic and non kinetic interceptors. AI and autonomy for automated decision making. Track fusion. Low latency communications to stitch it all together. Holmes described Golden Dome's ecosystem hub as a single front door for companies to find where they fit, with the primes subcontracting to nontraditional companies inside its command and control consortium, rather than the other way around.

Zach Martin

Map that to the three layers and you get a clean picture, and this mapping is mine. Infrastructure is the maneuverable vehicles and the interceptors. Distribution is track fusion and low latency communications, the connective tissue that decides whether a thousand sensors become one picture. And Applications is the mission itself, decision support for the people in the loop. What changes in autumn 2026 is that defense is no longer a customer that buys one layer from one prime. It's a customer asking for a stack, bought in pieces, on prizes. Romo's closing note is worth keeping. He said, quote, "We're swinging back pretty hard on manufacturing and building things again in America." Holmes called it a perfect storm, with capital markets, policy, and acquisition reform aligned at once. Storms pass. The question for allocators is who has the staying power when the headline cycle moves on.

Zach Martin

There's one more defense story, quieter than interceptors and in some ways more consequential for capital. It involves a program that got cancelled, and the money that was supposed to go to it going somewhere else entirely. That's a story about satellite navigation, and it's where we go next.

Zach Martin

Let's talk about a clock. Not a wall clock. The invisible kind. Every timestamp on a bank transfer, every millisecond in stock trading, every signal handoff between cell towers depends on GPS system time. Positioning, navigation, and timing, or PNT, is what Space Capital's research calls the foundation beneath a connected world. It's why your ride share driver finds your corner, why a combine can plant a field in straight lines, and why a financial network knows what order things happened in. And the system under all of it was designed in the 1970s for military navigation. Space Capital's write up on Xona Space Systems puts it plainly: GPS was never architected with the level of redundancy or resilience needed to support today's digital infrastructure.

Zach Martin

The mechanism of the weakness is worth a minute, because it explains where the money is going. GPS satellites sit about 20,000 kilometers above Earth. By the time their signals reach the ground they're weak. Buildings, terrain, and atmospheric interference degrade them, and in dense urban areas accuracy can slip by tens of meters. Worse, legacy navigation signals weren't designed to authenticate themselves, so a receiver has no reliable way to tell whether a transmission is genuine. That means spoofing, where someone broadcasts forged signals, can make aircraft misjudge altitude, ships drift off course, and networks run on falsified time. The research cites a 2019 NASA report of a passenger aircraft that nearly hit a mountainside in Idaho after a GPS disruption, with an alert air traffic controller stepping in. And in Ukraine, Russian jamming and spoofing have repeatedly disrupted drones and weapons. The system is a single point of failure that everything depends on.

Zach Martin

Here's the number that turns this from an engineering complaint into an investment thesis. Space Capital's research says enterprises spend over $20 billion a year on GPS hardware and augmentation services, not to unlock new capabilities, but to work around the system's limitations. Think about that. That's money spent on ground based correction networks, extra receivers, and backup systems, to patch a free public utility. The firm's argument, and I want to label it as their argument, is that this spending signals that the world is ready to pay for better PNT if a better system existed. The usual objection to premium navigation is that GPS is free. The counter is that being wrong isn't.

Zach Martin

Now the cancellation. According to Space Capital's research, the Space Force cancelled its billion dollar Resilient GPS program, known as R GPS, in early 2026, and pivoted toward contracting existing commercial low Earth orbit PNT providers. I'm relying on the research brief for that characterization, so take it with that attribution, and I haven't seen the underlying procurement documents. But if the account is right, think about what it means in the language of the three layers. A government program that would have built infrastructure itself is replaced by a purchase of a service from infrastructure someone else built. That's the same pattern we're about to see at NASA. The government stops being the builder and becomes the customer, and the market that results is one a private investor can actually underwrite.

Zach Martin

The company at the center of this is Xona Space Systems. It was founded by a group of eight engineers who first met in Stanford's Aero and Astro department more than a decade ago. CEO Brian Manning did structural design work on the Falcon 9 at SpaceX. CTO Dr. Tyler Reid earned his Ph.D. at Stanford's GPS Lab and then worked on localization for autonomous vehicles at Ford. The team reunited in 2019 around a conviction that legacy GPS wasn't built for a connected, autonomous, and contested world. And they looked at what Starlink did to broadband and what Planet did to Earth observation, two sectors once dominated by public agencies, and asked whether navigation could be next.

Zach Martin

Their answer is a constellation called PULSAR, a commercial high precision PNT service delivered from low Earth orbit. The physics is the elegant part. GPS satellites are large, government operated, and sit in medium Earth orbit. Xona's satellites are small and sit about twenty times closer to Earth. Closer means a stronger signal. Xona's own materials cite a signal 170 times stronger than GPS, and other summaries of the company put the figure around one hundred times. Either way, the claim is an order of magnitude or two, and stronger signals are harder to jam and work better in cities and indoors. The planned constellation is 258 small satellites. Let me be careful here: that's the plan, not the current state.

Zach Martin

Two other design choices matter for the economics. The first is timing. GPS satellites carry onboard atomic clocks, which are expensive and complex. PULSAR uses a patented distributed clock system that Xona says cuts satellite cost and complexity while keeping time synchronization accurate. That's the kind of detail that decides whether a constellation can be built at a price a commercial customer will pay. The second is compatibility. Xona engineered PULSAR to be a software based upgrade for existing GPS chipsets, so industries can adopt it without buying new hardware. And its signals are encrypted and authenticated, designed to resist spoofing. In investor language, that's a product with a low adoption barrier on the customer side and a defensible technical moat on the supply side.

Zach Martin

Now follow the money and the milestones. Xona launched its production demonstrator, Pulsar Zero, in June 2025, broadcasting what Space Capital describes as the first authenticated low Earth orbit navigation signal. In March 2026, following the successful in orbit validation, it closed a $170 million Series C to scale its factory in Burlingame, California. It has contracted with Aerospacelab to manufacture eight microsatellites for initial commercial coverage by 2027. And the customer list is already forming. Trimble is integrating for enterprise construction and agriculture corrections. The UK's National Physical Laboratory is a partner for distributing Coordinated Universal Time. And Xona signed memoranda of understanding with Furuno for industrial and marine timing and with Septentrio for drone autonomy. A memorandum of understanding is a statement of intent, not revenue. I'd want to see those convert.

Zach Martin

I want to be honest about the gap between the plan and the present, because that gap is where the risk lives. Eight satellites for initial coverage in 2027 versus a planned constellation of 258. A Series C that funds a factory ramp, but a long road to a global, resilient system. The sources I've got here describe Xona's approach and its financing, but they don't tell us how fast the Space Force or any other buyer will convert from cancelled program to signed commercial contract, or at what price. So when I say commercial PNT is replacing cancelled government programs, read it as direction and mechanism, not as a finished transition. The capital question is whether private money will fund the stretch between demonstration and a contract large enough to matter.

Zach Martin

Zoom out and look at what PNT does across layers. It's a Distribution story, because the product isn't the satellite, it's the signal moving through chipsets and software into someone else's product. That's why Trimble matters as a customer: it's an integrator that puts the signal into machines. And notice how it connects to the spectrum story from the first chapter. Space Capital's letter says Iridium's value to Rocket Lab lies in direct to device, aviation, and PNT, the markets heating up fastest. Put the pieces side by side. A launch company is buying an installed base of spectrum and subscribers in part because positioning and timing are becoming a paid service. A startup is building a new constellation to sell that service from lower orbit. And the government is reportedly choosing to buy it instead of building it. That's three different capital structures converging on one market.

Zach Martin

One more connection to the defense chapter, and it's a nice one. Romo's point was that high value orbits like medium Earth orbit, where GPS lives, are under pressure and can't be defended by sheer numbers. Xona's answer to the same vulnerability is architectural rather than defensive: don't just protect the one big constellation, add a second layer in a different orbit, with stronger signals and authentication. You can think of it as the proliferation philosophy applied to navigation, a different orbit and a different design so there's no single point of failure. Maneuver protects the crown jewels. Redundancy makes the crown jewels less singular. Both reduce the same risk, and both are reasons capital is moving.

Zach Martin

If I'm summarizing this chapter for an allocator, I'd say it this way. Positioning and timing used to be a utility you assumed. It's turning into a market you can underwrite, because there's measurable enterprise spend on workarounds, an identified government buyer shifting toward commercial supply, and at least one company with in orbit validation and fresh capital. What it doesn't have yet is scale, and scale is where these stories usually get tested. Which brings me to a company that's trying to change how you get to scale in the first place, by not owning the thing everyone assumes you must own.

Zach Martin

Thirty employees. No factory. Megaconstellations that customers own outright. That's the pitch Derek Huerta brought to the Space Capital stage, and it hit an assumption that nearly everyone in satellites has held since Starlink got big: that vertical integration is the source of the advantage. Eclipse Space came out of stealth on that stage. Space Capital led its seed round, so, same disclosure as before, this is a company the firm has backed. Huerta spent seven years at SpaceX as a satellite payload engineering manager, leading the team behind the phased arrays that link Starlink satellites to the ground. He spoke with Space Capital operating partner Tom Whayne, who was formerly CFO of OneWeb and chief strategy officer at Maxar, two weeks after SpaceX's IPO. Huerta said the technology is, quote, "too important to leave in the hands of just these trillion dollar companies."

Zach Martin

Roughly half of Eclipse's thirty employees came from Starlink, thirteen of them from the years the program was still called Satellite Development. They work out of Redmond, Washington, where SpaceX builds its Starlink satellites. First customer hardware ships later this year, and a demonstration spacecraft follows in 2027. Hold those dates in mind, because the claims are the company's own, and the proof is still ahead of us. Whayne opened with a scale problem that sets the stakes. In fifty years of satellite communications, only two companies have organically reached a billion dollars of connectivity revenue. Inmarsat needed thirty years. SpaceX needed, in his phrase, a handful.

Zach Martin

So what actually was the mechanism? Huerta named two things, and he said neither was vertical integration. The first was treating the user terminal, the spacecraft, and the launch vehicle as one system, so trade offs could be made across the whole thing instead of inside one slice. If a design choice in the antenna saves money in the satellite, you can make it. The second was importing mass manufacturing discipline from industries that had already solved it, and designing that discipline in from the start rather than retrofitting it later. His line was, quote, "You don't actually need to own a factory."

Zach Martin

That's the fabless argument, and Eclipse runs it the way Apple runs Foxconn and NVIDIA runs TSMC. Eclipse owns the design, the architecture, the integration, and the on orbit operations. Someone else owns the plant. And it engineers the spacecraft so that building it doesn't demand highly skilled labor, which is what lets a contract manufacturer do the work. Now here's the part that I thought was sharpest. Why has everybody copied vertical integration? Because a factory obligates you to fill it. SpaceX and Amazon generate their own demand, so for them the plant is always busy. Everyone else builds the plant first and then hunts for customers, and idle capacity raises unit costs, which prices out the buyers who might have filled it. The moat becomes an anchor. That's the phrase from Space Capital's write up, and I think it's the cleanest way to say it.

Zach Martin

Let me translate that into capital terms, with the label that it's my reading. A factory is fixed cost. If you build one, you need a financing structure that covers the fixed cost whether or not the orders show up, which means more equity, more debt, and more pressure to take any customer at any price. A fabless company turns fixed cost into variable cost. That changes who can finance the company and what kind of customer can afford the product. The upshot, in Space Capital's framing, is that constellation ownership comes within reach of buyers who could never build one: sovereign nations, regional telecom operators, and enterprises. I'll note that the full conversation covers the financing problem behind billion dollar constellations, and I haven't got the specifics, so I can't tell you how Eclipse solves it. I can tell you it's the question I'd ask first.

Zach Martin

Underneath all of it is one metric: dollars per throughput delivered in orbit. Drive that down far enough, Huerta argued, and serving a rural user from orbit costs less than serving them from a tower. At that crossover, the billions telcos spend every year on radios and towers redirect upward. Eclipse starts with direct to device because it's the hardest version of the problem, and the stack that solves it carries over to broadband and orbital compute. Notice how this lines up with the spectrum land grab in chapter one. If building a constellation gets cheaper and more accessible, the scarce asset isn't the satellite anymore. It's the spectrum the satellite talks over. That may be part of why Rocket Lab and Amazon paid what they paid for Iridium and Globalstar. That's my inference, not something either company said in the material I have.

Zach Martin

Then there's what Huerta called the third option, and it's geopolitical. He picked up a point made earlier that day by the NASA Deputy Administrator: space inverts the usual order, because commercial operators built the infrastructure and governments became the tenants. Countries are discovering they depend on capability they neither own nor control. China's answer, exported along the Belt and Road, is to build and finance your constellation and keep the strings. The American answer has been leased capacity on someone else's network. Huerta called Eclipse's model a third way: infrastructure built in country on what he described as an ITAR free platform, which the customer owns and operates. Those are his characterizations, and a sovereign buyer will test them hard. But the demand signal is clear. A government that wants to own, not rent, is a different kind of customer.

Zach Martin

My favorite detail in the whole write up is a small one. Before Eclipse had a logo, a website, or an office, Huerta had someone on a plane building the supply chain, and he had met heads of state. An advisor told him he was doing it backwards. The supply chain was the hard part, and the logo could wait. If you want to know whether a fabless company is real, that's the test. A fabless model isn't a model without manufacturing. It's a model in which the relationships and the integration are the manufacturing. It succeeds or fails on whether the supply chain actually delivers.

Zach Martin

Space Capital's own summary is that infrastructure gets built, becomes invisible, and then becomes the foundation of the next commercial layer: GPS, then geospatial intelligence, now satellite communications. Eclipse's bet is that this time the infrastructure layer itself becomes something you buy rather than a program you run. Keep that sentence in mind as we turn to the other big claim of the season, because the same logic runs through it, just at a much larger and more speculative scale. I'm talking about orbital data centers.

Zach Martin

The pitch is seductive: near limitless solar power, an effectively infinite heat sink, and not a single county permit. Elon Musk says all of it is going up. Jonny Dyer, CEO of Muon Space, put it with a wink at the Summit, saying it may be that Musk is right, and in three years all our data centers will be in space. He added that you may not be right about that timeline, but in some ways it doesn't matter, because the secular trajectory points there. Dyer sat down with Space Capital partner Tom Ingersoll, and Space Capital titled the piece Orbital Data Centers: The Honest Case, published August 20. Dyer's read: the physics works, the economics don't yet, and almost every obstacle in between is an engineering problem rather than a physics problem.

Zach Martin

Start with why it's attractive, because the physical advantages are real. The sun is stronger above the atmosphere, and in the right orbit it never sets. Dyer said you can pick orbits where the sun is baseload power, always on in a way you fundamentally can't get on Earth. Cooling is the second advantage: deep space is a nearly absolute zero background a data center can radiate into, without the water, the real estate, or the permitting fights that cooling triggers on the ground. And the third is regulatory. There are no county councils in orbit. If you come up with a regulatory approach, he said, it's a global approach, literally. The catch, in his words: capturing any of this is not physics, it's engineering.

Zach Martin

Now the honest part, the part that should keep an allocator's feet on the ground. At today's launch prices, a large orbital data center runs roughly eight to ten times more expensive than its terrestrial equivalent. Closing that gap means getting launch cost down about five times and cadence up about ten. Falcon 9 flies something like 100 to 200 times a year. The math, he said, wants 500 to 1,000. And he noted, luckily, that's also the scale of launch you need to put gigawatts of data centers in space, so the two numbers move together. Everything comes back to the cost of launch. And this is where the open uncertainty sits. The multi gigawatt story leans heavily on Starship reaching high cadence commercial operations, and that's an assumption, not a fact in evidence.

Zach Martin

Dyer then walked through the objections one by one, and what struck me is how many of them shrink under inspection. Communications first. Moving data between spacecraft is easy. The bottleneck is the regulated, bandwidth limited link down to the ground. But for AI workloads, most of the heavy traffic stays machine to machine. Starlink's inter satellite links run at about 100 gigabits per second, roughly an order of magnitude short of the terabit fabrics inside a data center. He called that a gap, not a wall. Radiation next. Terrestrial chips are already hardened against cosmic ray bit flips, because data centers hit those errors at scale on the ground. His favorite proof: Google's fleet effectively became, quote, "the world's largest cosmic ray telescope." And he said these problems are getting easier over time, not harder.

Zach Martin

What he refuses to wave away is that space is still hard, for reasons the easy button narrative skips. These are autonomous robots that have to take care of themselves, he said, and if you don't get that right, they will fail, and we can't go up and fix them. The deeper challenge is integration, stitching hardware, sensing, compute, and software into one system that works in orbit. Quote: "Integration is one of the hardest problems in technology, and we just haven't done it enough in space to get it to the point where it's muscle memory." That is an operator talking, not a salesman. And it's the same lesson as the fabless story: whoever owns integration owns the outcome.

Zach Martin

So none of it flips overnight. Dyer said we're not going to flip a switch tomorrow and suddenly have gigawatts in space. It's a progression. National security workloads, and data that's simply better kept in orbit, pull the first systems up early. Then cheaper launch and rack like satellites compound the effect. He pointed out that Starlink satellites look a lot more like a rack that goes into a data center than they look like a traditional satellite. And every hyperscaler will have to run the experiment regardless. He noted Google alone is spending something like $100 billion on data center capital expenditure this year, and a very small fraction of that goes a very long way above the atmosphere. That's his estimate and his framing, and I'd treat it that way.

Zach Martin

Here's the part that I think matters most for capital allocation. Near term, AI in space is delivering value through edge processing and sensor fusion on orbit, not hyperscale model training. Muon's FireSat constellation is the example. Built with Cal Fire, with backing from Google and the Environmental Defense Fund, it pairs orbital sensing with onboard AI compute and SpaceX optical links that Dyer said take a satellite from dial up to broadband. The goal is to give fire crews something they've never had: full situational awareness, overhead views of where the fire is, how hot it's burning, how fast it's moving, updated hourly, day and night. And Bloomberg reported that Muon raised $250 million with Google and Salesforce backing. That's real capital behind a real product that happens to be the first rung of the orbital compute ladder.

Zach Martin

Dyer's larger message was that the data center is the visible tip. He called it the ice above the water, with an iceberg underneath. For seventy years, he said, space behaved like the Galapagos, a separate ecosystem detached from the mainland of the economy, and that isolation is ending. Space and terrestrial technology will be so deeply integrated, in his words, that you won't even know the difference. And the corollary is a warning: those that don't have a strategy there are going to get left behind. Space Capital's own framing in the Q2 letter says the same thing from the investor side. Space is becoming AI infrastructure, and AI is becoming the economic engine of space. I covered the SpaceX version of that earlier this season, so I'll leave it there.

Zach Martin

Here's my synthesis of the chapter, labeled as mine. The fabless argument and the orbital compute argument are two halves of the same question, and the question is who bears the fixed cost of getting to orbit. If Eclipse is right, the satellite maker doesn't need a factory. If Dyer is right, the compute buyer doesn't need to wait for launch to be cheap to start learning. The Distribution layer, the networks and processing that move and extract value, is where both stories converge, because throughput per dollar in orbit is the metric on which both live or die. And the thing that sets the pace for all of it is launch cadence, which is why the entire sector keeps looking back at a rocket program in Texas. Now let's leave Earth orbit and go to the one place where the logistics are so basic they're just wheels.

Zach Martin

On December 14, 1972, Eugene Cernan drove the Lunar Roving Vehicle to a final position in the Taurus Littrow valley, parked it, and walked away. It stayed where he left it, alongside its two sister vehicles at Hadley Apennine and Descartes. The ascent stage couldn't afford the mass to bring them home. Their silver zinc batteries were single use. Their thermal systems were never designed for the roughly 300 degree swings between lunar day and lunar night. Across the three missions that used them, the rovers drove for about ten hours and covered roughly 90 kilometers. After Apollo 17, no American vehicle touched the lunar surface for the next fifty three years. That's the opening of Space Capital's August 5 piece on Lunar Outpost, and it's the cleanest way I know to explain why surface mobility is suddenly a capital story.

Zach Martin

What changed is policy and procurement. NASA Administrator Jared Isaacman announced the Ignition plan in March 2026, a ten year, three phase commitment to a permanent Moon Base. Under the plan, the orbital Gateway has been paused and the agency has reoriented toward what Isaacman calls operational urgency: faster timelines, commercial procurement wherever possible, and a willingness to take programmatic risk to get hardware on the surface within the decade. Artemis II flew its crewed lunar flyby in April 2026. Artemis III, slated for 2027, is a low Earth orbit rehearsal, and Artemis IV is targeting 2028 as the first crewed landing of the new era.

Zach Martin

Now let me bring in the demand signal from the government side, because it connects every chapter of this episode. NASA Deputy Administrator Matt Anderson, a month into the job, joined CNBC's Morgan Brennan at the Summit. His line was, quote, "We cannot do our job without commercial companies." The agency is now operating on hard dates set by national space policy: a crewed lunar landing by 2028, a commercial handoff of low Earth orbit by 2030. Anderson described the new contracting philosophy as incentivizing excellence. Deliver on cost, schedule, and performance and you get repeat business. Miss, and your contract gets canceled and the money redeployed. Schedule, which NASA treated as flexible for decades, is now non negotiable. And asked whether NASA prefers public or private companies, he said, "It's just outcomes. How you're funded is irrelevant to us."

Zach Martin

The architecture explains the urgency. Artemis III will put the three largest rockets in the world, SLS, Starship, and New Glenn, into orbit within days of each other for what Anderson summarized as three rockets, two dockings, and one splashdown. It's arguably the most complicated mission NASA has attempted. The destination for the landing is the lunar south pole, and the reason is water ice, because propellant produced on the Moon changes the economics of everything beyond it. Anderson was candid that early on the outpost, quote, "is going to look like a junkyard." I love that. It tells you NASA is describing an industrial buildout, not a highlight reel. And for low Earth orbit, the ISS retires in 2030, and NASA's requirement for the replacement is a commercial station where NASA is a major early customer but not the only one.

Zach Martin

This is the same pattern as Golden Dome's prize competitions and the Space Force's reported pivot on navigation. In Space Capital's words, NASA is converting itself from prime mover to platform: setting destinations, publishing dates, and paying for outcomes while private companies own the execution and, increasingly, the assets. For founders, the demand signal is legible for the first time in a generation. For investors, the largest customer in the market has told you exactly how it plans to spend. I'd add one interpretive note. A predictable, recurring customer is what makes a business model financeable. Anderson said the goal is a regular rhythm of awards so routine it stops being news, and that cadence of procurement might matter more to private capital than any single contract.

Zach Martin

Which brings us back to that vehicle parked in Times Square. On May 26, 2026, NASA selected Lunar Outpost and Astrolab to deliver the first human rated lunar terrain vehicles to the surface. Space Capital notes it was the first time NASA had selected commercial companies to provide astronaut surface mobility as a service rather than build it in house. The selection came as a $220 million firm fixed price task order to advance Pegasus, Lunar Outpost's lighter rover. The broader lunar terrain vehicle program carries a $4.6 billion ceiling through 2039, structured as per mission task orders rather than one fixed contract. That ceiling is a ceiling, not a commitment, and it's worth remembering that every task order is a separate decision. Another disclosure: Space Capital backed Lunar Outpost at the seed stage in 2022.

Zach Martin

Here's what Pegasus is, per the company. It carries two astronauts side by side in a low profile form that echoes the Apollo rover. It drives in three modes: crewed, teleoperated from Earth, and fully autonomous, at speeds of more than nine miles an hour across a 900 kilometer range, operating for a full year on the surface. It's compatible with both Blue Origin's Blue Moon Mark 1 and Starship landers. Two full scale functional prototypes have been through two rounds of suited human in the loop testing at Johnson Space Center with former NASA astronaut Dr. John Grunsfeld, and the company is charting a path to deliver the flight vehicle to NASA in November 2027. Eagle, the heavier concept, remains the long duration flagship. Pegasus gets there first.

Zach Martin

What I find most instructive is how the company got here. Justin and Julian Cyrus, the brothers who founded Lunar Outpost in 2017 with Forrest Meyen, both started their careers at Lockheed Martin and came away convinced that cost plus contracting rewarded delay and perfection over learning. So they built a small rover called MAPP and treated each build as a teacher. In 2021, NASA Administrator Bill Nelson presented the company a symbolic ten cent check, ten percent of its one dollar winning bid to collect a sample of lunar regolith, which Space Capital calls the first sale of space resources in human history. Then came Lunar Voyage 1, launched February 26, 2025 aboard the Intuitive Machines IM-2 mission. The Athena lander came to rest at an off nominal attitude on March 6, and the window closed before MAPP could roll onto the surface.

Zach Martin

Here's why that failure counts as data, and I want to be fair about the framing. Space Capital describes the mission as a partial success in the real lunar environment: MAPP validated its navigation computer, autonomous thermal control, stereo cameras, and a patent pending deployment system. It stood up the first commercial 4G and LTE network on the lunar surface with Nokia. And the Stargate command and control software reached technology readiness level nine, the highest NASA assigns. Notably, the rover didn't drive. So this is a company that earned flight heritage without achieving the headline outcome. You'd want to hold both facts. Lunar Voyage 2 is set for the second half of 2026 aboard IM-3, headed for the Reiner Gamma magnetic anomaly, and brings the first onboard GPU compute to the Moon with NVIDIA edge processors, so the rover can plan its own route rather than waiting on a signal from 384,000 kilometers away. The company says it has ten contracted missions before 2030.

Zach Martin

Now what the Cislunar Horizon means for capital. The argument Space Capital makes, and I find it persuasive though it is their argument, is that mobility is the layer that activates the rest. A power station no one can drive to is a stranded asset. A communications relay without endpoints is a tower in the dark. The 40 kilowatt fission surface power system targeted for 2030 arrives on a lander but reaches its site only by being towed across kilometers of terrain to a safe exclusion zone. Prospecting at Shackleton means traversing hundreds of kilometers of shadowed ground. Habitat shielding, sample collection, and relay installation all come down to moving mass between distant points. Mobility turns every other layer from a single landing into something that operates.

Zach Martin

Map that to the three layers once more. The rover is Infrastructure, physical assets on a celestial surface. The software suite is Distribution. Lunar Outpost describes Stargate as command and control, Starweave as swarm software that lets different robots operate as one fleet, and Spark as an AI mission operations assistant. A recent $30 million Series B is funding the teams building them. And NASA buying astronaut mobility as a service looks like an Applications layer customer with a first, very large order. The nearest test of that software is MARS One, which launches in fall 2026 aboard a Falcon 9 via an Exotrail transfer vehicle. Two small spacecraft will run decentralized swarm operations in low Earth orbit, the first spaceflight test of the engine behind Starweave. It was developed under Air Force Research Laboratory and Space Force SpaceWERX contracts, which is another place where defense dollars quietly fund a lunar product.

Zach Martin

The surface is also getting crowded. Space Capital counts six different programs with active lunar surface efforts, led by China and joined by India, Japan, South Korea, and the European Space Agency. China is developing a crewed rover for its 2030 landing, a six wheel design built for a single mission of roughly six surface hours. And the first Moon Base mission rides on Blue Origin's Blue Moon Mark 1 Endurance lander to the Shackleton Connecting Ridge, no earlier than fall 2026. That's another autumn milestone to watch. Space Capital's observation, which I'll quote loosely, is that a state run program builds one extraordinary rover for one mission, and a commercial market builds dozens, each refined against operational data. The firm cites Lunar Outpost's work for NASA, ESA, and the Australian Space Agency at once. In July, its Luxembourg entity was selected as thermal and systems engineering lead for ESA's Moonraker orbiter, launching around 2030.

Zach Martin

Let me pull the four stories together, because they're really one story told at four altitudes. In defense, the customer is shifting from buying a platform from a prime to paying prizes for demonstrated performance, and the scarce asset is maneuver. In satellite construction, the question is whether you need to own the plant at all, and the scarce asset is integration and, probably, spectrum. In navigation, a government is reportedly choosing to purchase a service from a commercial constellation rather than build a program, and the scarce asset is a trustworthy signal. In the cislunar economy, NASA is the anchor tenant of a commercial buildout, and the scarce asset is the ability to move things around. In every one of them, the government sets a date and publishes a budget, and a private company carries the execution risk in exchange for an outcome based payment. That's the industrial phase, as Chad Anderson called it.

Zach Martin

And here's the risk side, stated as plainly as I can. First, equity currency. Rocket Lab's Iridium deal doesn't close until around mid 2027, and the SpaceX Cursor deal was paid in its own stock, so these mega deals depend on post IPO sentiment holding up. Second, defense budgets. The proposed $1.5 trillion for fiscal 2027 isn't finalized, and prize based acquisition only works if the people running it keep running it that way. Third, launch cadence. Orbital compute and cislunar mass transport both lean on Starship reaching high cadence operations, and a rocket that isn't flying at scale yet is a very big assumption to hang a thesis on. Orbital data centers cost eight to ten times terrestrial ones today. That gap closes only if the launch math does.

Zach Martin

So here's what I'd watch as the fall unfolds, and I'm limiting myself to dates that the sources actually give. MARS One flying aboard a Falcon 9 this fall. Blue Moon Mark 1 Endurance heading to Shackleton no earlier than fall. Lunar Voyage 2 on IM-3 in the second half of the year. Eclipse's first customer hardware, due later this year. And beyond the calendar, the things Holmes and Romo said to watch: contract awards, test campaigns, and the quiet growth of space budgets. If those keep arriving, the rewiring is real. If they slip, we'll learn which stories were narrative and which were infrastructure.

Zach Martin

I'll leave you with the image Space Capital closes its Lunar Outpost piece on. Picture the South Pole by the end of the decade. A rover pulls away from a lander with an astronaut at the controls, tows a reactor to its site, then doubles back to haul regolith to a half built habitat wall, the ordinary rhythm of a working base. That's a vision, not a forecast, and I'd label it that way. But it's a useful one, because it reminds you that an economy isn't a flag. It's traffic. And the money this season, from Golden Dome to Redmond to Times Square, is flowing to whoever is building the roads. I'm Zach Martin. Thanks for spending the hour with me on SPACE CAPITAL, here on KMKT IR Hub Radio. Real assets. Real ideas. A brighter tomorrow.