In 1903, a deaf Russian schoolteacher named Konstantin Tsiolkovsky wrote down an equation and quietly ruined an entire industry before it existed. The math is not difficult. The consequences have been paid in dollars and Delaware filings for 120 years and counting.
This is Part 1 of a series on the space stack, told from the bottom up. The bottom is physics, and Tsiolkovsky is still down there, waiting for you.
The equation, stripped of romance
Here it is:
Δv = Isp × g₀ × ln(m₀ / mf)
Δv is how much your rocket can change its own velocity before running out of fuel. Low earth orbit requires 9.4 km/s of it.
Isp is specific impulse, a measure of how efficient your propellant is. You want this high. You are not going to get it much higher than it already is.
g₀ is Earth's surface gravity, hanging around for units reasons, like an uncle at a wedding.
m₀ / mf is the mass ratio: your rocket full of fuel, divided by your rocket empty. This is the number that hurts.
The cruelty lives in the natural log. Doubling your payload is a big deal. Doubling your Δv is a catastrophe. With a kerosene engine, reaching LEO requires a mass ratio of about 22 to 1 — twenty-two parts fuel, one part everything else. Engines. Tanks. Avionics. Payload. Fairings. The guy's lunch.
Pick a destination on the slider below and watch what Tsiolkovsky asks you to build.
Keep pushing the slider. You can feel the math start laughing at you. Past LEO, payload shrinks. Past GTO, structure is eating into what was supposed to be your satellite. Past Mars, in a single stage, the rocket is mostly tank and delusion.
This is why every rocket you have ever seen sheds pieces on the way up. Staging is not a clever engineering choice. Staging is Tsiolkovsky's tax, paid in hardware. You take sections of your rocket and throw them into the Atlantic so the parts you kept do not have to carry them any further. Each F-1 engine on a Saturn V cost several million dollars in then-year terms - and five of them splashed down about 560 km downrange in the Atlantic every Apollo mission. Nobody made a big deal about it.
This isn't abstract. Every rocket that has ever flown shows up the same way - mostly fuel, with the payload as a sliver on top.
A quick digression, because 9.4 is a weird number
Notice that 9.4 km/s figure. Low earth orbit is not actually moving at 9.4 km/s. Orbital velocity at 400 km altitude is closer to 7.8 km/s. The missing 1.6 km/s is not orbital speed. It is the bill gravity charges you for climbing up, plus a little drag and steering on top.
But there is a weirder physical fact lurking underneath, which people outside aerospace almost universally get wrong. Orbit is not "up." Nobody is in orbit because they are far from Earth. They are in orbit because they are moving sideways fast enough that when they fall, they miss.
Newton figured this out in 1687. Here is his version.
Newton gets you to orbit. Everything past orbit is transfer burns stacked on top of that first 9.4 km/s. The full bill, by destination:
So that's the bill. Nine point four kilometers per second, paid out of a tank with a 22:1 mass ratio, burned in stages you throw away. This is the physical landscape anyone trying to get to space has to cross. For most of the 20th century, only governments could afford it, which is a polite way of saying: only the people who don't have to show a balance sheet.
In the late 1990s, a group of very rich, very smart people decided that this was nonsense and that they could do it on their own. All of them were correct that the equation was payable by private industry. Almost all of them were wrong about the timing. Several are buried below.
The tuition: $200M and Andrew Beal's dignity
In 1997, a Texas billionaire named Andrew Beal decided he could build a rocket company without government help. Beal had made his fortune in banking and real estate. He had real engineering instincts. And he was not wrong about the core problem: launch was too expensive, the incumbents (Lockheed and Boeing, shortly to merge into United Launch Alliance) were slow, and private capital could theoretically fix both.
His rocket was called the BA-2 - a three-stage heavy-lift vehicle designed to put 17 tonnes into LEO. Its second-stage engine, the BA-810, was designed for 810,000 lbf of vacuum thrust - the largest liquid-fuel engine since the Saturn V's F-1. And Beal actually built it. On March 4, 2000, the BA-810 fired successfully at his test facility in McGregor, Texas. It was the largest liquid rocket engine ever tested by a private program, and the largest fired in America since Apollo. The first stage engine was bigger still, and never left the drawing board.
Three years later, SpaceX moved into the same McGregor site and turned it into its primary engine-testing facility. They rolled up and inherited the parking lot.
Technically, the BA-2 could have worked. Beal's engineers were real. The math was defensible. What killed Beal Aerospace was not physics. It was that the Air Force's Evolved Expendable Launch Vehicle program was funding Atlas V and Delta IV, guaranteeing launch contracts at prices that reflected thirty years of pre-absorbed taxpayer cost. Beal could not compete with a subsidy.
He announced the shutdown on October 23, 2000, in a public letter citing NASA's Space Launch Initiative - a $290M program backing competing reusable launch concepts - as the last straw. The letter was defensive, a little petty, and substantially correct about its narrow point.
The deeper lesson, in hindsight, is that Beal's framing was half-right. He was solving a hardware problem in a market where hardware was not the bottleneck. The bottleneck was capital structure - and also, as SpaceX would later prove, the way you work with government customers, not against them. A self-funded billionaire cannot absorb ten years of negative operating income while a competitor flies on public money. He needed patient outside capital and a government anchor customer, and in 2000 neither was available to a rocket startup.
He was not alone down there. The late 1990s were a rich time for smart, well-funded, technically credible private launch companies that all died of the same thing. Click a stone.
Everyone in this graveyard died of the same disease. The symptoms looked different — rotors, spaceplanes, hypergolics, exotic composites — but the cause of death was always capital structure, not physics. For forty years after Apollo, nobody fixed it. Nobody even tried that hard.
Then Elon got bored at PayPal.
Before we get to what SpaceX actually did differently, one chart. What it cost to put a kilogram in orbit, across every major launch system of the space age.
A note on the numbers: launch cost accounting is slippery, especially for government programs where development and overhead get amortized creatively. Figures below are directionally right, individual numbers are contested, and Saturn V alone ranges 10x across sources depending on methodology. Primary sources are Harry W. Jones (NASA Ames, 2018) and CSIS Aerospace Security. Details in the chart footnote.
Two things jump out. For forty years, launch costs didn't meaningfully go down. The Shuttle — sold to Congress as the reusable spacecraft that would finally make space routine — turned out to be more expensive per kilogram than the Saturn V it replaced. This is not a rounding error. This is an entire generation of engineers optimizing in the wrong direction while Congress applauded.
Then, around 2010, something broke loose. And it was not a new equation.
The unlock: nobody beat physics, they beat Excel
SpaceX did not beat the rocket equation. Nobody has. The Merlin's Isp is roughly on par with the F-1's. Raptor is about as good as hydrolox without being hydrolox. Mass fractions on Falcon 9 are impressive but not physics-defying. What SpaceX beat was the capital stack around the equation, and the industrial process underneath it.
The engineering unlock was reusability with minimal refurbishment, which is importantly not the same as reusability. The Shuttle was reusable. Every orbiter came back and needed thousands of hand-hours between flights: heat-shield tiles inspected and replaced one at a time, SRBs fished out of the ocean and repacked, main engines torn down and rebuilt. The Shuttle program recovered its hardware and then spent enough money putting it back together that it would have been cheaper to throw it away. Which is roughly what they eventually did.
Landing a Falcon 9 booster vertically and reflying it in three to four weeks with mostly inspections and a few component swaps is new. Flying the same booster thirty-plus times is new. The relevant cost saving is not "the rocket is reusable." It is the rocket is a capital asset you depreciate over many flights instead of an expense you write off on every mission. That is a balance-sheet innovation. Tsiolkovsky does not care about balance sheets.
The capital-stack unlock is the part where Beal's analysis was actually wrong. Beal argued government support of competitors made private launch impossible. SpaceX proved the opposite: you succeed by becoming the government's vendor, not by beating it. Musk stacked a DARPA Falcon 1 contract, a NASA COTS award (2006), the CRS cargo contract (2008), and eventually an anchor tenant he founded himself (Starlink, which needed the rockets to exist). That math works. The lesson was not "take on the government." It was "get the government to pay you."
Why this matters for everything above
When $/kg falls 20x, the entire stack above it changes shape. Satellites that used to be impossible become merely expensive. Satellites that used to be expensive become commodity. Constellations that were bankruptcy machines in 1999 become investable businesses. Missions that required a decade of political cover become Tuesday's manifest.
This is the single most important fact about the space sector today, and it is the reason every other layer we are about to tour got funded. Ground stations got interesting because the thing in the sky got cheap. Earth observation got interesting because you could afford to put up more than one camera. In-space servicing got interesting because someone might eventually fly your wrench.
Andrew Beal wasn't wrong. He was early, which in venture capital is the same thing as wrong, just more dignified about it. The tuition he paid, along with everyone else rotting in the graveyard above, is part of what made the next twenty years cheap enough to matter. Tsiolkovsky is still down there. But he is finally being paid in volume discounts.
Next: Part 2, The Vehicle. If you can afford the equation, the next question is what to build. The answer, for a surprising number of companies, was "strap a rocket to a 747." That worked about as well as you'd expect.