What this covers
This is a wide-ranging primer in which Patrick draws out Tren Griffin on how the space industry works, how it got here, and where it might go. It starts with a map of the economy: space-to-Earth, Earth-to-space, and a newer space-to-space segment. It then turns to history, where cost-plus contracting is blamed for bloated programs, since contractors had little reason to cut launch costs. Against that sits Elon Musk's approach: question the assumption that rockets are thrown away, land them, use commercial rather than radiation-hardened parts, and iterate in public through frequent failures. Griffin puts commercial launch at roughly $25,000 to $30,000 per kilogram before Falcon 9 and about $2,500 after.
The middle of the conversation is technical and commercial. It covers why Starship's size improves efficiency, how cubesats opened participation beyond defense contractors, and why sub-scale launch providers may not survive. A long stretch covers wireless physics: the Spectrum Trade-off between penetration and bandwidth, the capacity limits of LEO cells compared with a GEO satellite, and the failure of 1990s satellite phones that tried to copy cellular voice.
The more contested claims come later. Griffin argues that Starlink's best profits will come from services layered on connectivity, that defense increasingly depends on ubiquitous low-latency broadband, and that Musk's price-elasticity bet remains unproven. He also treats the $30 to $60 billion Mars price tag as a question of priorities, not affordability. The episode closes on governance, where space can't be owned and the ITU regulates without enforcement, and on a power-law view of innovation in which failure is expected.
Griffin argues the space economy's value sits mainly on the ground, in services and connectivity that satellites enable, and that falling launch costs, driven by reusability and demand elasticity, will expand that base even though launch itself is a low-margin business.
- Ground equipment and services (about $260B combined) dwarf the launch market (about $6B), so space enables more value than it directly generates
- Lower launch prices create new demand through elasticity, and reusability and iteration drive costs down in a compounding feedback loop
- Satellite networks win where they serve moving devices and places terrestrial networks cannot reach, not by replicating fiber or home broadband
Connectivity and direct-to-consumer services, not bandwidth, are satellite's largest opportunity and Starlink's value.
- As chips, software, and networks become more central, ubiquitous satellite coverage becomes more valuable because it lets connected devices work everywhere.
“we have chips in everything and those chips run software and they're all connected by networks ... space enables that because it's ubiquitous in terms of its ability to give coverage everywhere”
- Big LEO satellite phone systems tried to replicate cellular voice service and failed because they were expensive and clunky and did not work in cars or houses, so users had to stand in a field.
“they said, 'Okay, well, we're going to create a cell phone that works everywhere.' ... 'big LEO satellite phones are great for people who are out standing in a field.'”
- The largest satellite opportunity is connecting devices rather than people, such as cars that receive software updates and services through antenna tiles built into their bodies.
“I think the biggest opportunity is less connecting people and more connecting devices. ... those tiles can connect to networks in space”
- A dish antenna mounted on a moving car does not work in practice because of physics, so the viral Starlink-on-a-car stunt was staged, and future flat electronically steered panels are needed for mobile use.
“I think that was all staged, but there's no way it worked. The Dishy doesn't work on the move. It just can't, it's physics.”
- Starlink will be most profitable from services built on top of connectivity rather than from selling bandwidth itself, just as internet providers earn modest margins while applications on top capture the most value.
“I think Starlink will be most likely to be a gusher of cash, not from selling bandwidth, but from selling services on top of that bandwidth.”
- A LEO constellation covers every location over time, so rural and urban users get the same coverage, but each cell has limited capacity, so providers must manage oversubscription or risk degraded service.
“you can't oversell that because if you oversell that, you oversubscribe. If you oversubscribe, we could have bad service”
- Multiple LEO operators such as OneWeb and SpaceX will persist regardless of which business wins, and some externality from earlier investments, like abandoned subsea fiber, becomes useful later for cloud connectivity.
“it was a bust, but those cables were around for us to say, Hey, let's connect data centers and create cloud connectivity.”
- Musk chose a Tesla-style direct-to-consumer model for Starlink with no dealers, selling services and terminals straight to users who can use them out of the box, unlike enterprise-heavy competitors.
“I'm going to be like Tesla. I'm going to have no dealers. I'm going to go to direct to consumer”
- Reviews that judge Starlink against fiber miss the point, because satellite connectivity offers different value through mobility and ubiquity, and Musk calls it better than nothing rather than a fiber replacement.
“Of course it isn't better than fiber. But it's different than fiber.”
Cheap launch comes from reuse, iteration through failures, vehicle scaling, and marginal pricing of unused capacity.
- Under cost-plus government contracting, contractors resisted lowering launch costs because cheaper launches would reduce their revenue, which produced bloated programs and dysfunction.
“the way to make money was to basically be a cost plus contractor ... the contractors thought, 'if we lower the cost of getting people to space, we will make less money'”
- Musk's Starship aims to cut launch cost toward about $20 per kilogram to the Moon, compared with about $12,500 per kilogram for a Falcon 9 to low Earth orbit.
“he talks about taking the cost of launch down to $20 a kilogram to the moon, because the $20,000, or the $12,500 on a Falcon 9, just gets you up to low Earth orbit”
- Musk examined the launch industry's assumptions and realized that discarding rockets after one use made cheap access impossible, so making rockets that land attacked the core cost driver.
“the fundamental one was, you can't have a good business if every time you drive to the store, your car ends up in the ocean and then you can't use it again.”
- Machine learning applied to data from each launch lets engineers see fixes that would not occur to people, so every landing improves faster and reusability lowers cost further.
“you have a lot of data and you have machine learning and you're getting all this data from every launch ... It's like the computer's saying, this is obvious.”
- Frequent launches generate data that accelerates learning, which improves the product and enables more launches, creating a feedback loop that Falcon has entered and that did not exist before in space.
“the more times you launch, the more times you learn, the more times you learn, the better your product gets, the more times you launch. You get into this feedback loop”
- Commercial launch cost was about $25,000 to $30,000 per kilogram before SpaceX, and Falcon 9 brought it down to about $2,500 per kilogram to low Earth orbit by reusing rockets and changing assumptions.
“it was $30,000, $25,000 a kilogram ... And so, they now got it down to $2,500 per kilogram to LEO.”
- Volume lowers unit cost and improves product quality through data and customers, which is why sub-scale launch providers risk exit, as LG's phone exit illustrated.
“volume means everything. If you're not making 30 million phones, you're out, like LG just dropped out ... the more you make, the cheaper it gets”
- Starship is more efficient than smaller rockets because enlarging a vessel improves efficiency geometrically, much as bigger oil tanks cost less per unit of volume.
“as they expand the size of a vessel that you're going to put stuff in, you basically get more efficient. It's the same reason why oil tanks are so big in Oklahoma”
- SpaceX's strategy of developing in public and iterating through frequent failures, including blowing up rockets, is a cheap way to develop low-cost rockets because each test produces data that improves the next design.
“we're going to develop in public. And we're going to have this iterative development where we're blowing up rockets all the time.”
- SpaceX can price unused launch capacity down to marginal cost, like an airline selling empty seats, which allows cheap rides for student experiments, ashes, and other secondary payloads.
“He can price that thing down to marginal cost like an airline.”
- If Musk's Mars goal requires about $10 per kilogram, his approach of pricing each cost reduction to marginal cost will keep pushing launch prices down, which is only sustainable because Starlink generates cash.
“if Elon really wants to go to Mars, he needs it to be $10 a kilogram.”
- SpaceX's rideshare flight carrying about 88 satellites, with room for extra Starlink satellites, shows how cheap launch lets many small experiments in messaging and imaging reach orbit.
“he took this one rideshare program up, where he had like 88 satellites on it, and all kinds of experiments are going on”
- Even if the launch market doubles from $6 billion to $12 billion, that is small relative to the $130 billion ground services market, so financing big projects must come from services rather than launch revenue.
“even if the launch market increases from $6 billion to $12 billion, it's peanuts in terms of financing a mission to Mars”
Failure is inherent, and decentralized experimentation and embracing outlandish bets beat conservative, centralized space practice.
- Decentralized teams empowered to change designs produce step-function improvements, whereas centralized rule-setting that rejects savings because they are not in the spec stalls progress.
“'I could save 10% doing this.' No, you can't do that. That's not in the spec. ... Having a decentralized system is what allowed the Falcon”
- Space ventures follow a power-law pattern in which most bets fail and a few deliver outsized returns, so failure is inherent to innovation and the sector should expect it.
“There's always a non-linear return. And you're always going to have a power law of some things aren't going to work and some things are. ... Failure is inherent in innovation”
- Boeing's defense and space president said that of 30,000 engineers, only seven could understand and integrate an entire airplane's design, financing, and resources, so integrative system thinkers are the scarce resource.
“I got seven that can design an entire airplane and do the whole thing, 30,000 engineers. ... only seven can understand the whole thing and how it fits together”
- Humans will likely not live on the Moon in an Antarctica-style setting by 2035, with about 2060 a rough guess, but once progress arrives it will be larger than expected.
“I would say, I'll just SWAG, more like 2060. There's a lot of things that have to be done for it to go right.”
- Traditional space engineering prizes proven, radiation-hardened technology, which adds cost and risk aversion, whereas SpaceX chose the latest commercial technology, reduced costs, and iterated in public through frequent failures.
“The traditional assumption in building satellites and building things for space is very conservative. ... Musk said, 'No, we're going to take the latest technology, we're going to reduce costs.'”
- Technological progress tends to be slower than expected in the short run but larger in the long run because it is non-linear, as with the internet and software.
“progress tends to be slower than you think in the short run, but in the long run it's bigger ... Things tend to go non-linear, progress tends to go non-linear.”
- The traditional space industry was conservative, using proven, radiation-hardened technology that gradually moved from ground to space, and that conservatism added cost, which contractors were paid for under cost-plus contracts.
“The traditional assumption in building satellites and building things for space is very conservative. Can't take any risk.”
- The best bets are half-crazy: ideas that are not completely implausible but look outlandish to the mainstream, because they contain optionality that conventional projects lack.
“you don't want it to be completely crazy, you want it to be half crazy.”
- Space investing is attractive because the outcomes are unknown, and if the outcome were known it would be boring, so uncertainty itself is part of the appeal.
“this is the funnest game on earth. We don't know the outcome. If we knew the outcome, it'd be boring.”
- Future vehicles will carry flat, electronically steered antenna tiles built into their skin, using metamaterial technology, allowing every device to connect to satellites, a direction Griffin says he is investing in personally.
“eventually there will be an electronic or an antenna that will look like a panel.”
- The F-35 communication system uses every band from S-band up to millimeter wave and is embedded in the aircraft skin, and this type of technology is expected to spread to drones and other platforms as it gets cheaper.
“Its communication system uses every band, S all the way up to millimeter wave. It's embedded in the skin.”
- Most of the projected trillion-dollar space economy will come from new services that have not yet been conceived or were not expected to be large.
“most of that 1 trillion will be new services that we haven't even thought about today”
Ground-side markets and devices, not launch, hold most of space's economic value.
- The space economy consists of space-to-Earth and Earth-to-space activity, plus a newer space-to-space segment, and most of the economic value happens on the ground.
“the key thing to understand about the space economy is it's two things, which is, it's space to Earth and Earth to space. ... And the interesting thing about the space economy is most of it happens on the ground”
- The space economy is roughly a $6 billion launch market, a $13 to $14 billion satellite manufacturing market, about $130 billion in services, and about $130 billion in ground equipment, so ground-side markets are far larger than launch.
“you have approximately today a $6 billion launch market. And then you have this $13 to $14 billion satellite manufacturing market. ... services, which is $130 billion or so and then ground equipment, which is another $130 billion”
- Morgan Stanley's forecast of a one trillion dollar space economy by 2040, with about half from broadband, is a rough guess rather than a reliable estimate.
“Morgan Stanley says the space economy is growing to one trillion in 2040, which is a big guess. It's a SWAG.”
- A single satellite is launched once and serves millions of devices on the ground, so spending on ground devices and services far exceeds spending on the satellite itself.
“You put a satellite up there and you've got one satellite, and you'll launch it once ... But you have millions of devices talking to that satellite. Obviously, you have a lot more spending going on on the ground.”
Cheap launch and mobile satellite services must grow because terrestrial networks cannot serve their needs.
- Starlink lets SpaceX act as its own launch customer, filling unused rocket capacity at near marginal cost, which reduces the demand uncertainty that is its biggest financial risk.
“one of the ways to create enough demand for launch is to be your own demand. And so there's Starlink ... He can price that thing down to marginal cost like an airline.”
- SpaceX is best understood as a services company with an attached launch business, because launch is important mainly as the enabler of services that terrestrial networks cannot provide.
“to me, Space X is a services company that has an attached launch business. And it's an important business, that launch business, because it's an enables the services.”
- If Falcon and Starship keep cutting prices toward marginal cost, from $2,500 toward $500 and then $200 per kilogram, competitors built on high-cost state-funded models would be unable to keep up.
“I'll take it down to $1,500. I'll take it down to $500 with Starship. I'll take it down to $200. ... your product is some state supported thing that has thousands of dollars of costs”
- Launch-enabled satellite services must keep getting cheaper because they compete with 5G and fiber, and they need uses terrestrial networks cannot serve, such as GPS-like and remote messaging services.
“because they're competing against 5G. They're competing against fiber, but it's got to get a lot cheaper. And then also there have to be new uses, which terrestrial can't do.”
- Small standardized cubesats, measured in units, can be dispensed in large numbers and carry experiments cheaply, breaking the old model in which only defense contractors building large satellite buses could participate.
“it used to be, you had this giant thing that was called the bus ... And the defense contractors made it ... along came some people ... cubesats”
- Musk identifies the biggest risk to SpaceX as financial, namely not running out of cash, and the most uncertain variable is whether enough launch demand will exist.
“'What's our biggest risk? Our biggest risk is financial.' We've got to basically not run out of cash.”
- A LEO constellation covers New York and remote Africa equally, so the network is a blanket of coverage that must be sold to airlines, ships, and everyone else to fill capacity, not only to dense markets.
“you have the same coverage in New York city that you do in the most rural place in Africa ... you basically have this blanket of coverage and you have to sell it to everybody.”
- Space companies should not simply replicate ground services, since a mobile, nomadic human workforce wants connectivity that works while moving, as Craig McCaw recognized when he designed wireless around mobility rather than office phones.
“don't just replicate what's on the ground ... You have to think create new value. ... I think most of that 1 trillion will be new services that we haven't even thought about today”
- The biggest threat to the space market is terrestrial alternatives, which can limit its size, and critics who compare Starlink unfavorably with fiber miss that satellite service is different and designed for other situations.
“The biggest challenge for space is what's happening on earth. ... Of course it isn't better than fiber. But it's different than fiber.”
Autonomous systems and defense needs make space connectivity more valuable than ever.
- Chinese AI assisting pilots in dogfights points toward a next step where fighter jets have no pilots, because pilots are expensive to train and cannot react as quickly as machines.
“There was just an article about the Chinese having AI that helps their pilots to be better dog fighters. The next step is just no pilots.”
- Modern defense is shifting toward small, mobile special forces, autonomous drones, and machine-learning-driven systems that all depend on ubiquitous low-latency broadband, making connectivity more valuable to defense than ever.
“we're basically moving to a defense force that's more and more special forces ... machine learning fighting machine learning. More and more autonomous drones.”
- Defense depends on satellite communications across many bands, and advanced embedded antennas on platforms like the F-35 show why connectivity is essential, even though such systems are expensive today and getting cheaper.
“its communication system uses every band, S all the way up to millimeter wave. It's embedded in the skin. It's expensive, but that kind of technology is getting cheaper.”
- Space lacks electrical outlets so satellites must be solar powered, and automated robotic repair vehicles that attach to satellites can extend their life and add power, as already demonstrated on a GEO satellite.
“there are no electrical outlets out there. And so everything has to be solar powered ... You have automated repair. ... it attaches to your satellite and gives you more power or fixes this”
Every orbit and frequency band involves unavoidable performance trade-offs.
- Low radio frequencies penetrate walls and water but carry little bandwidth, while high frequencies carry much more data but need line of sight and degrade through obstacles, so every wireless design trades one property for the other.
“You have really low frequencies, and the good news about them is they go a long way and they go through walls ... But as you get higher, it doesn't go through stuff anymore.”
- Geostationary orbit sits about 36,000 kilometers above the equator, lets a satellite stay fixed relative to Earth, covers roughly a third of the planet, and offers only a limited number of slots.
“there's only one geostationary orbit, which is 36,000 kilometers in the sky, over the equator ... you can see a third of the Earth, basically”
- Lower orbits are cheaper to reach because less energy is needed to lift mass, while higher orbits such as geostationary cover more area but cost more to reach, so each orbit involves trade-offs.
“the lower you are, the less it costs to move, because you have to use energy to move mass from the Earth. So it's cheaper to put things in a low earth orbit”
- Modern software-defined chips let networks adaptively choose whichever frequency works in a given situation, reducing the penalty that high-frequency bands face from poor penetration.
“with software chips today, you can have adaptive networks that use the frequency that works”
- Satellites will never be a great trunking technology because they need many more hertz per bit than fiber, so fiber will remain better for high-volume data transport.
“satellite will never be a great trunking technology just because it needs so many bits for hertz, and fiber's better.”
- LEO constellations have finite capacity per cell, so overselling subscriptions in a dense area would degrade service, while a single new GEO satellite can offer about a terabit per second.
“you can't oversell that because if you over sell that, you oversubscribe ... For a really big GEO, the new ones going up will have a terabyte per second of capacity.”
Exceptional people and integrated dreamer-doer teams drive progress, and talent clusters around them.
- Space programs need chemists, physicists, electrical engineers, software developers, regulators, and business people integrated together, and people who can quickly absorb new disciplines, as Musk is described as doing, have an advantage.
“every discipline's involved, you've got chemists, you've got physicists, you've got electrical engineers”
- SpaceX attracts extraordinary people, and talent clusters because extraordinary people like to work with other extraordinary people, creating a self-reinforcing talent flywheel.
“the thing about life is extraordinary people like to be around extraordinary people.”
- Exceptional doers may be products of circumstance, and their skills create value only when placed in the right situation, as Buffett suggested that his skills would have been worthless in another era.
“It's like Buffett says sometimes, if he'd been born 300 years before, he'd be nothing.”
- Geoffrey West argued that cities thrive because they attract unusual people, some of whom occasionally produce breakthroughs.
“Geoffery West at Santa Fe said this great thing about why cities are so amazing. Because they have crazy people.”
- Societies need both dreamers who pursue audacious goals and practical doers who build things, because the doers turn ideas into working systems and the dreamers provide direction.
“we need dreamers and doers. The thing about some of these people is they do shit. They don't talk about it.”
- Gwynne Shotwell complements Musk's dreamer role by handling business and numbers and making sure operations run on schedule, and SpaceX's ability to recruit extraordinary people is essential to its progress.
“Elon's a dreamer, but she's going to make sure the trains are all on time.”
Cheaper launch may unlock unforeseen uses, though demand elasticity remains unproven.
- Musk's bet is that cutting launch prices from about $20,000 per kilogram to about $2,500 will sharply increase demand, a price-elasticity thesis that remains unproven.
“he believes if you take the price of launch from $20,000 ... down to $2,500, people are going to launch more stuff and they're going to find new uses. ... the model hasn't been proven yet”
- After Sputnik, the United States devoted a huge share of GDP, over one percent at its peak, to catching up with the Soviets and reaching the moon.
“we devoted a huge portion of GDP to basically catching up with Sputnik and putting people on the moon ... it's over one percent of the economy at its peak”
- In 1993, Griffin and Russ Daggatt proposed a broadband LEO system using Ka-band spectrum with about 840 satellites, which others dismissed as crazy but which Griffin saw as the kind of half-crazy bet that carries optionality.
“we said 840 satellites in 1993. When we did that, people said, 'Well, you're just nuts.' ... you don't want it to be completely crazy, you want it to be half crazy.”
- Lowering the price of a good often increases use in unpredictable ways, as with Bill Gates's radial tire example, so cheaper launch will enable many uses that cannot yet be foreseen.
“Sometimes like Bill Gates famously said about radial tires, sometimes just people spend a lot less money and they drive around a lot more, but it enables other things. Lower price launch is going to enable a lot. We just don't know what it is.”
- Some pharmaceutical and electronics products are better made in zero gravity because the environment provides advantages, which is a more straightforward near-term case for space manufacturing than moving heavy industry off Earth.
“there's some things in zero gravity that are better to make, just the environment is better for that. And it can be some pharmaceutical thing, some electronic thing.”
Ambitious space ventures carry large societal value, from inspiring young people to moving humanity beyond Earth.
- Although space carries trade-offs and needs careful rules, its potential for good is much larger than its potential for harm.
“The potential for good though is bigger than the potential for bad. Everything has trade offs, but the potential for this to be wonderful is much, much bigger”
- Ambitious space projects inspire young people, including those from disadvantaged backgrounds, to pursue science and engineering, creating a positive externality that benefits society beyond the projects themselves.
“the dreamers have a positive externality of inspiring everybody. ... You're in some tough family situation ... 'I want to go to space. Like, that's cool.'”
- Musk's motivation for SpaceX is to move human consciousness beyond Earth, paralleling Tesla's goal of electrifying the automobile to improve the planet.
“his motivation for Space X was to move consciousness beyond earth. ... his goal with Tesla was basically to make earth a better place by electrifying the automobile.”
- The money needed for Musk's Mars ambitions, roughly $30 to $60 billion, is small relative to the global economy and the available capital, so it is a question of priorities rather than affordability.
“he's going to spend $30 billion to do that or 60 or whatever it is. That's not that much money, not in terms of the global economy.”
- Jeff Bezos's idea of keeping Earth residential and moving polluting manufacturing into space is a radical but serious vision, though it faces physics challenges that lower launch prices would need to overcome.
“what Bezos is talking about, which is t zone earth, residential and put all polluting manufacturing in space. ... I think there's physics challenges.”
Space governance and spectrum rest on fragile treaties that no authority can enforce.
- Satellite spectrum allocations such as Starlink's were won by a small team at a World Radio Conference, and remain tenuous because any country can refuse to follow the rules.
“Basically a team of six of us at the start. Then eventually about 25 people got that allocation ... at the World Radio Conference ... it's always tenuous because somebody can just say, 'Nope, I'm not going to follow the rules.'”
- Space cannot be owned as property because the frame of reference rotates with Earth, so governance rests on treaties and the ITU, which issues regulations but has no enforcement police force.
“nobody can own space ... So they came to this conclusion, well, everybody owns space. ... The ITU ... they don't have a police force. They can't enforce anything.”
Mars is a leap requiring full convoy logistics and self-sufficient survival, not an incremental step.
- Mars is aligned with Earth only about every 26 months, so Musk's plan requires assembling a convoy of around a hundred Starships in orbit and launching them together when the window opens.
“Mars is only aligned with the earth at a certain time. It's every 26 months or some number. Elon's plan is you take a lot of these Starships, like a hundred of them”
- Going from the Moon to Mars is a leap rather than an incremental step, because Mars requires solving the convoy logistics, habitability, and sustained operations that the Moon does not.
“But I think from moon to Mars isn't just a step. It's a leap.”
- Mars travel requires carrying everything needed for survival, so SpaceX's head Gwynne Shotwell described it as the ultimate camping expedition, and early settlers will need to be geologists who can make materials on site.
“she just said, 'It's the ultimate camping expedition,' because you've got to bring everything. ... Some of the early people who are going to go to Mars are going to be geologists”