Ten thousand generations of our ancestors watched a red star brighten as it chaotically reversed its course across the sky. Within living memory, our technological ascendancy saw that star become a world, a distant twin of Earth. Long ago, rivers fed lakes and oceans and, perhaps, life teemed in pools now long since dried up. Mankind has dreamed of distant shores, strange vistas, expansion across an endless frontier. Now, for the first time, Mars is within our reach. How do we get there?

Mars is 100 million miles away, so we’re going to need an enormous rocket. Ideally, thousands of them. The SpaceX Starship was built for this singular purpose, the hopes and dreams of the only known consciousness riding a crackling pillar of fire to open the heavens for our infinite descendants, expanding throughout the universe.

This technology has matured at the same time as terrestrial politics turn towards the heavens once more. Donald J. Trump became the first president to explicitly set going to Mars as a goal of his second term in his inaugural address, proclaiming: “we will pursue our manifest destiny into the stars, launching American astronauts to plant the Stars and Stripes on the planet Mars.”

The SpaceX Starship’s design encapsulates the challenges of getting to Mars, getting back, and transporting vast quantities of cargo to humanity’s first city on another planet. Along the way, Starship radically redefines our ability to launch satellites and scientific instruments in a way that threatens the very existence of our legacy space organizations.

How Starship Will Transport Us To And From Mars

Going from Earth to Mars requires several distinct steps: a short but violent launch into Earth orbit, then a long and peaceful glide to the other planet, followed by a short but violent re-entry and landing on alien ground. In the words of Robert A. Heinlein, “once you get to Earth orbit, you’re halfway to anywhere in the solar system.” Rocket launches are spectacular because they must muster the power necessary to defeat the gravity of the entire Earth. After ten minutes or so, the rocket has accelerated the payload to the incredible speed of 17,000 miles per hour (mph). Once in orbit, the payload will stay there indefinitely, clear of Earth’s atmosphere and its drag.

But low Earth orbit (LEO) is just barely outside Earth’s atmosphere. To get to Mars, we need an additional boost to escape Earth’s gravity altogether. At certain times, Mars’ location is just right that a relatively small additional boost puts our Starship on a trajectory that, about eight months later, results in us running into Mars. This is the tried and true method used by our intrepid orbiters, landers, rovers, and Mars helicopters. These launch windows open up for about eight weeks roughly every two years—2.14 years to be exact—the synodic period of Earth and Mars. For example, the next optimal short transfer launch date is November 2026. Then, December 2028, February 2031, April 2033, and so on, with transit times taking between 100 to 150 days through to 2050.

Launching to LEO and then coasting to Mars is non-trivial, but landing on Mars is super tough. First, the Starship arrives at about 8 kilometers per second (km/s), far too much speed to negate with rockets. Instead, we use Mars’ tenuous atmosphere to slow us down. Mars’ atmosphere is 150 times thinner than Earth’s, but it’s just thick enough to catch a speeding Starship, loft it above the highest mountains, then slow it down to about the speed of sound. While NASA’s various landers and rovers used a parachute and skycrane to slow down, Starship is far too large for that. Therefore, it employs its flaps to generate lift to the lowest possible speed, then fires its engines at the last second for a controlled soft landing on the surface.

No matter how much cargo we send to Mars, eventually we’re going to want to be able to return cargo, rocks, and people. Getting back from Mars is much harder than getting there. Earth has breathable air, rocket factories, and an unlimited supply of rocket fuel. Mars has spiky rocks and total ambivalence about our continued existence. The one point in our favor is that Mars is considerably smaller than the Earth, requiring much less velocity to reach orbit and with no atmospheric drag losses either. While Starship is designed to be able to fly from Mars to Earth on a single tank of gas, it arrives on Mars with nothing left in the tank. We need a source of fuel on Mars.

The easiest way to get fuel on Mars is to import it from Earth. As of early 2025, Starship is designed to transport 100 metric tons of cargo to Mars. Starship requires 1200 metric tons of fuel and liquid oxygen to fill. Therefore, assuming zero losses and some kind of low overhead piping system on Mars, twelve cargo flights bringing fuel to the surface is enough to send one Starship back to Earth.

There are also a few tricks we can employ to reduce the number of flights. For example, we don’t need about half the fuel until we’re in low Mars orbit (LMO), so rather than shipping that fuel all the way to the surface we can park a tanker from Earth in LMO and perform ship-to-ship refilling (or crew transfer) on the way home. Mars’ atmosphere is 95% carbon dioxide (CO₂). By mass, 80% of Starship’s propellant is oxygen (O₂). With a relatively simple albeit electricity-intensive chemical process we can ingest Mars’ atmosphere, strip off the oxygen, and store it cryogenically on the surface, reducing the number of fuel shipments needed from twelve to three.

If we’ve solved the problem of generating abundant electricity on Mars, it’s even possible to produce the methane (CH₄) fuel Starship needs on Mars from atmospheric carbon dioxide and water, via the Sabatier process. It may even make sense to produce the fuel locally from day one. This will represent the early stages of a fully independent industrial economy on Mars. The relative ease of producing methane and oxygen fuel on Mars is the reason Starship uses this fuel combination.

Like the trip to Mars, the trip back to Earth will be shortest roughly every two years. As an Australian, it amuses me that the time taken for physically plausible journeys to and from Mars on a chemically-propelled monster rocket like Starship is similar to the time taken for my ancestors to sail from Great Britain to the colonies about 150 years ago. Unlike them, though, voyagers in Starship will not have to endure hunger, disease, infant mortality, and navigational errors. While landing on Mars has yet to occur, landing back on Earth will not be a problem, as demonstrated recently by the incredible “chopstick” maneuvers of Starship test flights 5 and 7.

Starship’s harsh mission requirements have informed its design. The rocket is crazy light and has the world’s lightest, cheapest, most powerful, and most efficient rocket engine. Its engine design is so ambitious it has never before been attempted, even with hypergolic propellants, even in the Soviet Union which was known for its closed cycle rocket innovation. To not just reach Mars but settle it, it is necessary also to manufacture these tremendous rockets in the hundreds and thousands. SpaceX has already built a one million square foot facility in Boca Chica, Texas to mass-produce Starships.

All That Has Come Before is as Nothing Next to Starship

As I write this, SpaceX has just landed their 400th Falcon rocket. No other operational rocket has been reflown. Europe, India, Russia, Japan, and China are still flying old school single-use rockets with minuscule cargo capacity, low flight rate, and high operational cost. A number of U.S.-based private companies including Rocket Lab, Blue Origin, and Stoke are developing their own reusable launch vehicles, but they have some catching up to do. Falcon has done an admirable job of launching anything and everything to space. In 2023, humans launched around 1.5 million kilograms to space—80% on Falcon 9.

The majority of the most recent launches are serving SpaceX’s own Starlink internet satellite constellation. In 2019, I predicted that Starlink would be an enormously profitable venture, providing excellent service to millions of people around the world. Today, the team at SpaceX has made this a reality, earning nearly $8 billion from nearly 5 million customers in 2024, with 269% growth year-over-year. Starlink currently operates about seven thousand satellites, more than triple the total number of operational satellites just a few years ago. Starlink provides the torrent of gold needed to continue funding development of Starship and its cargo for Mars. Each bit of information transmitted by Starlink produces enough profit to fund the transport of a picogram—about a trillion atoms—of cargo to Mars. A million tonnes of cargo on Mars will cost 10²⁴ Starlink bits, equivalent to total Internet traffic in 2024. 

But where Falcon is designed to launch millions of kilograms per year, Starship is designed to launch a million tonnes per year, a thousandfold increase over Falcon. Falcon is for constellations. Starship is for nation building. For space people of my generation, raised on a diet of necessarily pragmatic and minimalist bootstrapped space colonization ideas, Starship’s firehose of mass funded by Starlink’s river of gold seems almost a bit excessive.

Yet we continue to grossly underestimate the difficulty of the enterprise. Beating Factorio is not even a warm-up compared to challenge of launching a million people to Mars, then fitting them into an industrial stack that can operate in a pitiless hostile irradiated vacuum and also maintain one-hundred times the productivity of contemporary industrialized nations, none of which can even approach industrial autarky here, on our home planet, with fewer than fifty million people.

One thing we can know for sure is that the ultimate productivity of the early settlers on Mars will face a long line of constraints. To the extent that these constraints can be addressed on Earth, where air is free and labor markets deep, we should absolutely do that! What is the marginal return to the enterprise of performing mass optimization on every machine we have to send to Mars, versus buying ten standard commercial machines off the shelf and turning the Starship factory up to eleven? As SpaceX’s Principal Mars Development Engineer Paul Wooster has remarked, “you can hide a lot of sins with upmass.” I would go further and say that lifting the upmass cargo constraint is a necessary, but not sufficient, condition for success at the Mars colonization effort.

Ideally, we’d spend zero effort optimizing or constraining our cargo manifest. The ideal Starship is a transparent logistics system that moves stuff from Earth to Mars like a routine, predictable conveyor belt. Earth-Mars launch windows force these shipments to be stockpiled in Earth orbit and launched en masse every 2.14 years. Continuous launches of fully and rapidly reusable fuel tanker Starships and Super Heavy boosters lift a million tonnes of fuel and liquid oxygen into orbit. There are 8760 hours in a year, so this lift will require a launch roughly every hour. 

Roughly ten tanker launches are required to fully refuel a cargo Starship in LEO, so every ten hours a cargo Starship is launched, refilled in LEO, then placed in a parking orbit. Refilling cargo ships in orbit allows us to send ten times more cargo per Starship than if we launched directly to Mars from Earth’s surface with a single stack. In essence, refilling adds a “free” third stage to the rocket.

After two years, two thousand fully-fueled cargo Starships are ready and waiting in orbit. Most of these will fly one-way to Mars, since it’s much easier to make a new Starship in the giant shiny optimized Starship factory here on Earth than to refuel and return a Starship from Mars. Some relatively small fraction of the Starships will have crew compartments, and, in the days before departure, their crews will arrive, ferried up from Earth. An even smaller fraction of these Starships will be intended to one day return to Earth with travelers, scientific samples, and cargo. 

While the Earth-Mars launch window opens for about eight weeks, the optimal transfer time is less than two weeks. During this time, departure trans-Mars injection burns can occur above the Earth on the opposite side of the departure vector, which is to say they will be visible nearly continuously in the tropical evening skies. On average, every ten minutes a parking orbit will intersect the departure plane, light its engines, burn for about ten minutes across the sky, and rapidly depart as its plume remains lit by the setting Sun. 

Roughly six months later, two thousand Starships will loft over the horizon on Mars burning an incandescent trail through the sparse atmosphere. Crewed ships will arrive first on faster trajectories, followed by cargo ships. Orbital mechanics dictates a landing in mid-afternoon, and unlike the departure burn, landings must be synchronized with the planet’s rotation so that we land at the Mars base, rather than some random location on the planet. If we space landings out over a one hour window per day over six weeks, these periods will see roughly one landing per minute. The spaceport will be busy!

When You Have a Starship Hammer, Everything Looks Like a Nail

While Starship was built to transport a million tonnes of cargo to Mars, its exceptional capabilities are about to cannibalize the rest of the space industry, including ultimately Falcon. The good news is launch capacity will increase by 1000x while costs will fall. The bad news is that none of the legacy aerospace companies or organizations are ready for this. They must either increase productivity by a hundred or thousand times, or else be lapped by newcomers who will.

Russia’s Roscosmos recently flew the 2000th flight of a Soyuz rocket, part of the R-7 rocket family that first flew in 1957. Expendable with a measly payload of just 8 metric tons, it’s obsolete. France’s Arianespace has sat on their hands for a decade, recently launching Ariane 6, obsolete years before delivery. China has an active launch program, but all their active launchers are similarly antique. At least they have vigorous public-private development of Falcon 9-class launchers, but as far as I know, nothing in the Starship category. Back home in the U.S., Blue Origin recently launched New Glenn for the first time, a rocket comparable in performance to Falcon Heavy, which first flew in 2018.

Starship has flown seven full-stack test flights thus far, and is soon to launch its first payloads. Each test has demonstrated new capabilities and design updates as SpaceX pushes towards full operational capacity of Starship with full reuse of both the booster and upper stage.

Conventional wisdom in space engineering holds that space vehicles must deal with a bewildering variety of constraints. Power constraints. Thermal constraints. Data transfer constraints. Material constraints. Above all, mass constraints. Space is hard, rockets are extremely non-negotiable about payload capacity, and there’s never quite enough launch capacity to saturate demand.

This reality drives exceptional space technology development organizations, such as NASA’s Jet Propulsion Laboratory, to laboriously design and construct exquisite spacecraft, such as the Mars Perseverance rover. This nuclear-powered exploration robot, which weighs about a tonne, had component masses tracked during design down to 0.1 grams! It took years to build, cost over $2 billion, and entire components were carved out of chunks of solid titanium. Despite recent layoffs, the laboratory still has thousands of staff who specialize in mass-optimized spacecraft design. What is the point of spending years and billions of dollars on shaving grams off a Mars robot when Starship can deliver one hundred tonnes of anything to Mars for less than $100 million? 

The superpower of Starship is replacing a menagerie of expensive limited bespoke vehicles for tentative, programmatically brittle human Mars exploration with a single, unified, powerful robust system. While legacy contractors and national agencies have been relatively slow to recognize and capitalize on the capabilities Starship brings online, private industry and new space contractors have stepped up. Better, larger satellites and better, larger space telescopes—including telescopes and interferometers large enough to view distant extrasolar planets orbiting other stars—are only the beginning. The possibilities are manifold:

Building New Space Stations and Habitats

It costs NASA about $4 million per person per day to operate the International Space Station (ISS), a price so high that private industry has proven unable to justify any kind of extended use of space facilities for research. Elite science test facilities such as synchrotrons are just $50,000 per person per day, making them a relative bargain. Laboriously constructing, launching, and operating some modular space station in the style of the ISS over thirty years is not a model that can reduce costs by 100x. 

What if, instead, we used Starship to launch a large, self-contained space station with supplies pre-loaded? With stations rolling off a production line, design improvements and customizations are easy to integrate in a controlled environment here on Earth. Once supplies are depleted, the orbiting assets can be safely splashed in the Pacific Ocean or potentially sold on to other operators. The startup Vast was founded a few years ago to test this hypothesis. This is the path to find an economically sustainable model for deep space habitation, which can scale up to much larger habitats that rotate to produce artificial gravity.

Exploring the Moon

Starship’s cargo capacity is transformational in terms of lunar exploration, a far cry from the cramped landers of the Apollo program. Each Starship can serve as a self-contained base for extended exploration or, in a more ambitious program, replicate the Mars colonization program and convey thousands or millions of tonnes of cargo to the surface. Starship makes something like the Antarctic program possible on the Moon, with dozens if not hundreds of engineers and scientists living at the south poles of both the Earth and the Moon! A number of new companies have already begun building pieces of this program, including Venturi Astrolab, who are building a versatile Moon rover.

This newfound capacity is both an opportunity and a risk for incumbents. Starship blows NASA’s expensive Space Launch System (SLS) out of the water. While SLS can launch a modest payload, Starship is designed to be able to go all the way, land there, and then return to Earth—a complete capability within a single system. NASA is incredibly fortunate that private industry has come to the table with the Starship capability at this time. Business as usual without SpaceX would almost certainly see NASA begging China for a ride to a lunar base for decades to come.

Predicting the Weather

Starlink already flies seven thousand highly capable communications satellites around the Earth. Starlink signals can already be used to perform a role comparable to civilian GPS, and could be improved further. The Starlink antennas form dozens of beams to transmit data across multiple frequencies to millions of customers all over the world. Each of those beams uses modern coding protocols which are robust to noise and signal loss. Starlink can record real-time frequency-keyed beam opacity data for each of its links, then feed this data into a real-time data engine to perform high resolution global atmospheric tomography. Obtaining a trillion daily measurements of atmospheric temperature and humidity at all altitudes is a dream dataset for weather prediction. When combined with emerging improvements in AI, I expect to see a drastic improvement in forecasting, perhaps weeks into the future.

Starlink ships with multiple highly capable antennas. With minor modifications, this sort of hardware can perform both active and passive multistatic synthetic aperture radar (SAR). In principle, this can also gather the weather data mentioned above, in addition to gathering high resolution 3D ground imagery independent of lighting and clouds. Because radar reflection is mostly a function of surface electron density, SAR can measure soil moisture, intertidal inundation, snow depth, as well as the built environment.

Umbra is a startup capable of delivering state-of-the-art SAR images with just eight small satellites. Imagine what a partnership with Starlink could deliver! A Starlink-based radar capability can also be turned outwards, greatly increasing the capabilities and resolution of planetary radar used to detect and track asteroids that may threaten the Earth. High quality timing data can also enable the entire constellation to function as a single networked passive antenna for radio astronomy and the search for extraterrestrial intelligence.

Launching Probes, Rovers, and Miners

Traditionally, the high cost of launch drove high spacecraft costs and infrequent flights in a reinforcing doom loop. While our technology has greatly improved, the rate of development for new space science missions has slowed to a crawl relative to the days of Lunar Ranger, Surveyor, and the Mariner programs, which launched dozens of space probes. Partly, this is driven by top heavy space science divisions where single minded “principal investigators” must push a mission concept for decades before it flies, usually when they’re on the cusp of retirement if not death. There are no second chances. Even if the mission is successful, the hard-won expertise of project leadership is often lost each time as senior leadership retires rather than ploughing their knowledge back into a continuous exploration campaign.

Starship provides an opportunity to disrupt this paradigm. What if, instead of building ever more expensive, ever less frequent behemoth one-off missions, we flew a Starship’s worth of science instruments to every planet at every opportunity? NASA and their partners must adapt their mission development processes towards a production line, enabling continuous design improvement and real-world testing. Mishaps are expected and priced in. Schedule slippage is no longer catastrophic nor particularly expensive, as cargo slots can be freely bought and sold between a much larger ecosystem of collaborative science instrument developers.

Alleviating transport capacity to the Moon, Mars, or moons of Jupiter will rationalize the economics of science capability development, promoting a hardware-rich mentality. As we gear up to launch humans back to the Moon and to Mars, we desperately need abundant geological survey data for prospective landing locations. Starship enables us to easily launch probes to every planet, every year.

Starship’s launch capacity opens the door to two other opportunities. The first is asteroid mining. We can bombard prospects with generous assay equipment, mining robotics, and ore processing machinery—or deflect unwanted space rocks that threaten the Earth. The second is a rapid response to recently-detected interstellar visitors. In 2017, we detected ‘Oumuamua, a speedy interstellar object, as it was on its way out of the solar system. Because we were unprepared, we had no ability to chase this mysterious object and have a closer look. Now it is gone forever. It could be an extrasolar asteroid or a piece of destroyed alien spacecraft, but we have no way of knowing. Now that we know these things exist, we have no excuse to not be ready for the next one.

The Core Technologies for Settling Mars

Until Starship has matured, it makes little sense for SpaceX to devote any of its finite development capital to technologies needed to live on the surface of another planet. That means lots of new ground for old companies, new companies, and you to explore. The opportunity is to find business models that support dual-use technology development for a terrestrial market, then fund operations for space. 

We need to find ways to deliver industrial quantities of electrical power on Mars. Solar and nuclear are the two obvious ways. The advantage of solar is that it is inert, has no moving parts, and relies on essentially panels strewn at random over the landscape. The disadvantage is that production is affected by weather, seasons, and diurnal cycles, requiring batteries for load shifting at night. Mars is also further from the Sun than the Earth. It’s fair to assume that power consumption on Mars for both life support and industrial processes will be far more even than on Earth, where demand swings widely depending on the time of day. Nuclear power is obviously far more complicated, but the problem statement is at least simple.

While importing fuel to return to Earth is possible, it will likely turn out to be cheaper to set up enormous hydrocarbon synthesis plants on Mars to consume yet more power and stockpile methane and oxygen. Similar plants can also produce other hydrocarbons essential for the petrochemical industry. We’re unlikely to find naturally-occurring oil on Mars, so if we want the ability to make plastics, paints, dyes, glues, chemicals, pesticides, fertilizers, and medicines, we’ll need to produce all kinds of synthetic oils. (In my day job I develop similar technology for application here on Earth. We’re hiring!)

We need to mine air, water, and rocks on Mars. The air miner ingests the Martian atmosphere (cold, sparse, mostly CO₂) and separates it into different gas products, including trace quantities of water vapor (H₂O). We can also readily produce oxygen from carbon dioxide, meaning we can use only an air miner to maintain the atmosphere of our habitat on Mars—a useful trick! Serious industry requires more water than an air miner could produce. While occasionally we see evidence for water on the Martian surface, it’s thought to be more common at depth, in the form of permafrost, buried glaciers, or in some rock-ice mix. In some areas, sufficiently deep, there may even be geothermally heated liquid water. 

Sooner or later, we’ll need to localize production of cement, steel, aluminum, and other metals. It is highly unlikely that our landing place will have, in addition to appropriate properties for landing rockets and copious solar power and water ice, concentrated and large ore bodies for the one or two dozen most commonly-used metals. Therefore, we’ll need the ability to set up nearly autonomous remote mining outposts where ores do exist, and/or the ability to process undifferentiated basaltic rock, which Mars has in abundance, to separate out the metals we need. This is chemically easy to do, but extremely power intensive. There is a cost and complexity trade-off between trying to manufacture certain products on Mars versus importing them from Earth via Starship. To a good approximation, goods that currently travel by air will be imported, while goods that travel by ship may be eventually locally produced.

Robust, reliable, long-duration life support systems are an unsolved but extremely worthy problem. Just for starters, all the plumbing has to work in a variety of different gravities, including Earth, Mars, the Moon, and the weightlessness during cruise in deep space. Life support machinery on the ISS routinely breaks down for all kinds of annoying and banal reasons. For example, clogged filters, failed impellers, and catalysts poisoned by atmospheric contaminants. Space suits for Mars are also an unsolved problem, as are communications and navigation systems; we will probably get to set up a Starlink constellation in Mars orbit just for quick communication.

The first people to live on Mars will live in their landed Starships or others shipped for the purpose. But if we’re going to do much of anything, we need to obliterate the constraints of a hostile environment and minimal space. There’s nothing romantic about attempting to replicate the entire industrial stack while hot-bunking on the equivalent of a WWII submarine. The most effective way to create a shirt-sleeves environment over an area of Mars measured in tens of square miles—the sort of area required to stand up dozens of mostly automated factories and farms—is to tent it with a thin flexible membrane.

Anchored to the ground every hundred feet or so to provide pressure relief, a kevlar-reinforced transparent shell can provide a warm, breathable, natural-light environment with effectively unlimited ceiling height. This material is similar to a high-performance yacht sail, and could even be manufactured on Mars with reasonably rudimentary processes. 

Conventional views of construction on Mars see astronauts in space suits swinging pick axes and moving dirt with wheelbarrows. This approach has not scaled on Earth for more than a century and will not work on Mars. The good news is that numerous heavy machinery manufacturers on Earth already produce a wide range of excavation equipment designed to work in hostile environments and are hard at work on remote operation and automation. The opportunity lies in finding ways to customize this supply chain to work not at the bottom of the ocean or deep in some mine, but in the frozen near-vacuum of the Martian surface. This will require a vacuum-compatible hydraulic powerpack and vacuum-compatible lubricants, paints, and electric insulation. 

Robotic construction machinery can be operated remotely, which is an advantage given that labor on Mars, even pushing buttons on giant diggers, will be extremely scarce. The challenge is that Mars is 4 to 25 minutes of light-time away from Earth, incurring a messaging delay. NASA has decades of experience uplinking sequences enabling piecemeal automated execution of days or even weeks of work for the rovers. On Mars, some combination of remote control, local control, and full automation will be necessary, but this capability is also enormously valuable here on Earth and still largely undeveloped.

In addition to technology that can convert rocks into metals and plastics, we will need factories to convert these raw materials into useful products. There is a well-trodden path of national industrialization that typically begins with labor-intensive manufacturing such as textiles, and then proceeds up the value chain until it ultimately outsources all physical manufacturing altogether, for better or for worse. 

On Mars, we will need to speed-run this timeline. We won’t have the labor or logistical capacity to plug into an existing trade network covering a billion people, so we’ll have to be able to make bulk products and critical necessities, over time diversifying into a wider range of products. To achieve anything like industrial autarky with only a million people, we will need to increase per capita productivity by a factor of at least one hundred, equivalent to all the productivity advances since the beginning of the Industrial Revolution.

Starship is the bridge to another world. Every test flight gets us closer to full operational capability, even as production continues to ramp. In our lifetimes, thousands of Starships will carry millions of tonnes of cargo and millions of people to a new world. What are you going to do on Mars?