The long road to Mars: Starship needs 1,000

I spent a weekend with a spreadsheet, a physics textbook, and the SpaceX Starship specs. I wanted to understand the logistics of getting humans to Mars. Not the “wouldn’t it be cool” version. The math version.

The numbers are staggering.

The basic problem

Mars and Earth align for efficient transfers roughly every 26 months. During each transfer window (which lasts a few weeks), you can launch spacecraft on minimum-energy trajectories that take about 6-9 months to reach Mars.

SpaceX’s plan requires sending approximately 1,000 tons of cargo to Mars per transfer window to establish a self-sustaining colony. Starship can carry roughly 100-150 tons to low Earth orbit, but the payload to Mars is much less (because you need fuel for the transit and landing).

With orbital refueling (launching tanker Starships to fill the Mars-bound ship in orbit), each Mars-bound Starship might deliver 100+ tons to the Martian surface. So you need about 10 fully loaded Mars-bound Starships per window.

Each Mars-bound Starship needs roughly 6-8 tanker launches to fuel it in orbit. That’s 60-80 tanker launches per window, plus the 10 Mars-bound launches, plus the launches for crew ships.

Roughly 100 Starship launches per transfer window. Every 26 months.

For a mature colony program, the numbers get bigger. A lot bigger. SpaceX has talked about 1,000 ships to Mars. The launch cadence required for that is… not something the current infrastructure can support.

Where we are

SpaceX has launched Starship about 5 times total. The turnaround between flights has been months. The rapid reusability that makes this math work (launching the same booster multiple times per day) hasn’t been demonstrated yet.

The gap between 5 flights total and 100 flights per window is enormous. It’s the difference between a prototype and an airline.

Why I don’t dismiss it

Because I’ve seen this before.

In 2015, SpaceX had never landed a rocket. By 2024, they’ve landed over 300 boosters. The Falcon 9 launches roughly once every 3-4 days. What was impossible became routine in under a decade.

Starship is a bigger vehicle, a more complex mission, and a harder problem. But the methodology is the same: build, fly, break, learn, fly again. The iteration speed is the secret weapon. Not the size of the rocket. The speed of learning.

Could SpaceX go from 5 flights to 100 flights per window in 10 years? Five years ago I would have said no. After watching them catch a booster with a tower on the first attempt, I’m not saying no anymore.

The supply chain nobody talks about

Even if SpaceX can launch often enough, what goes on those ships? A Mars colony needs everything. Food production systems, water extraction equipment, power generation, construction materials, medical supplies, spare parts, computing infrastructure, radiation shielding.

And people. People who are willing to spend 6-9 months in a steel tube, land on a planet with no breathable atmosphere, and build a civilization from scratch. With a two-year wait for the next supply shipment.

I tried listing every supply category a Mars colony would need. I got to 200+ before I stopped. Each category requires its own supply chain, its own manufacturing capability, its own maintenance plan. A colony that can’t produce food locally runs out of food when a ship doesn’t arrive. A colony that can’t manufacture spare parts goes dark when a critical system fails.

Self-sufficiency on Mars isn’t a technical challenge. It’s a civilizational one. It requires reproducing, in miniature, most of the industrial capability that Earth took centuries to develop.

The timeline

SpaceX says boots on Mars by the late 2020s. NASA says 2040s. The Planetary Society says “it depends on how much money Congress appropriates.”

My personal estimate: first humans on Mars by 2035, give or take 5 years. First permanent settlement by 2050, if everything goes well. Self-sustaining colony by 2080, if we’re lucky.

These timelines feel too long to some people and too short to others. That’s probably a sign they’re in the right range.

The math is daunting. The logistics are almost incomprehensible. But so was “reuse a rocket” in 2010, and I watched a rocket get caught by a tower this year.

The spreadsheet says it’s insane. The progress curve says it’s possible. I’m holding both thoughts at the same time and trying to figure out which one to trust.

Related thinking: