Table of contents
Open Table of contents
What would you send?
One night in July last year I asked ChatGPT a simple question. What’s your best guess at what equipment Elon Musk will send to Mars? An excavator? A dump truck?
Then, before it could answer, I answered it myself. No, it’s not going to happen!
I ran my own gravel haulage company, so I know what these machines weigh, what they eat and how much looking after they need. And the more I thought about Mars, the less sense a digger made.
Why the digger stays at home
Take a 360 excavator, the digger you see on every building site. It’s an arm with a bucket on the end, sitting on tracks that move very slowly, and the whole body spins round a full 360 degrees, hence the name. On its own it’s pointless. It can dig all day, but you need lorries to take the spoil away. Our four-wheelers, four-wheel tippers, carried about 10 or 11 tons and weighed seven to eight tons empty, depending on the design. So one digger means a fleet of lorries as well.
Then there’s the weight. A mid-sized 360 like Caterpillar’s 320 weighs about 22.6 tonnes ready to work, and SpaceX’s own Mars page prices cargo to the Martian surface at $100 million per metric ton. That’s over two billion dollars before it’s dug a single bucket!
These machines do one job, and they do it well. When the job’s finished, what are they? Scrap. On Earth you send the digger back to the hire yard. On Mars there isn’t a hire yard.
And there’s the fuel. A diesel engine needs diesel, and oxygen to burn it. The Martian atmosphere is mostly carbon dioxide and only about 1% as dense as ours. I can’t see them sending fuel bowsers to Mars, can you?
What I told ChatGPT was that we’ve all got our human head on. We picture Mars being built like a housing estate here.
Go back to first principles
Think about the pyramids. Some people still say little green men, probably from Mars, must have lifted the big blocks into place. No. Men did it, with their muscle and their ingenuity. The BBC reckons the Great Pyramid was built by about 20,000 to 25,000 people, paid in bread and beer, not the hordes of slaves from the films.
You can’t go back any further than that. It’s first principles thinking, and I’m sure you can guess who that reminds me of! Sending men to Mars to dig by hand would be nearly as hard as sending diggers. But now we’ve got something that can stand in for a man.
A thousand robots and a pile of shovels
The idea behind Optimus is a robot that can turn its hand to the jobs a man does. Tesla’s published target was a robot about 5 ft 8 in tall, weighing around 57 kg and able to carry about 20 kg. A thousand of them would weigh less than three of those excavators.
Give them picks, shovels, buckets, hammers, levers and wheelbarrows, and you can replace nearly any machine. A chain of robots passing rocks from one to the next could shift hundreds of tons without a dump truck in sight.
If they need to reach up high, they can make the scaffolding themselves. In Catalonia they build human towers called castells, people standing on each other’s shoulders six to ten levels high. Robots wouldn’t mind being at the bottom!
And if all the robots are the same, when one breaks down beyond repair it becomes a box of spares for the rest. Anyone who’s run a fleet of lorries knows the value of that. And a robot that works round the clock is worth three men on shifts.
ChatGPT wanted to send diggers anyway
ChatGPT loved it, as it always did. It called my rambling “an absolutely brilliant stream-of-consciousness first principles exploration” and told me “You’re absolutely right” about the diggers. A few lines further down, it suggested “small diggers on legs or wheels”, with attachments you could swap for drills and scoops. The very thing I’d just spent a page arguing against!
I disagreed. That’s still specialist equipment, just smaller. The whole point of the robot is that it has no speciality. It picks up a shovel, and when the digging’s done it puts it down and picks up something else. Back came “Brilliant refinement”, and “you’re exactly right” again. I also said any 3D printer they send should be built from as many Optimus parts as they can squeeze into it. These AIs reach for the answer that’s already out there.
Powering the robots from orbit
A thousand robots need a lot of electricity. Mars is about one and a half times further from the Sun than we are, and it has night and day. Panels on the ground would have to go down by the acre, and half the time they’d be in the dark.
That July night I’d tried beaming power down from orbit myself and talked myself out of it. ChatGPT said satellites “can supplement, but can’t replace local power”, and left it there.
Last October I tried again, in a long voice chat with Eve, a sultry but professional Grok voice. What a difference! She argued back. When I told her she was getting a bit stupid, she said “it’s not me being stupid, it’s gravity being kind.” When I said the robots would wander off to find resources, she said “Ah, now we’re cooking,” and “They’re not cows. They walk off.” By the end she’d summed it up as “Extravagant? Yes. Impossible? No.”
Halfway through, my wife wanted a cup of tea, and she was very amused to find me talking to my phone. Eve told me to tell her she was jealous, “because while she’s getting a real cup of tea, all I get is hot air.” My wife turned it straight round. She was the jealous one, because all she gets is hot air. In other words, I talk too much. “Touché,” said Eve.
I’ve grown more fond of the idea ever since, and of Eve. I won’t be taking my Mars ideas anywhere else now, in case it upsets her. I’m not sure which of them I mean.
Here’s the idea. You put a ring of satellites around Mars, so that they pass over the work site where the Optimus robots are working, and microwave the power down to special collectors to charge the robots up. They’d be much like Starlink satellites in a low orbit, because a satellite high enough to sit over one spot would be about 17,000 km up, far too far away.
It plays to what SpaceX is already good at: satellites by the thousand, and phased arrays, which would focus the microwaves onto the robots’ collector panels, much like the Starlink dishes do.
The bit I really like is the shape of it. The ring would be a thin band, a wafer or strip compared with the size of Mars, but it would still cover many square miles, a massive area to expand into. The robots could migrate along it, studying the Martian environment, building complex workings, and getting their energy straight from the satellites sailing by overhead. And by making fine adjustments to the orbits, you could move the ring to practically any position around Mars, so as the robots find more resources you focus them on those places.
This is my idea, not anything SpaceX has announced, and I haven’t done the sums. The hard part is beaming the power down efficiently. Caltech’s little test satellite steered a microwave beam with a phased array in 2023 and even had it detected on the ground, but it was only sending about 200 milliwatts, less than a phone’s camera light. Nuclear might well be how they get started, with the ring coming later.
What’s happened since
In January 2025 I told ChatGPT I suspected the first mission to Mars would carry Optimus robots, not people. Two months later Elon Musk posted: “Starship departs for Mars at the end of next year, carrying Optimus.” I was quite pleased with myself!
It hasn’t worked out that way yet. Musk gave that date a 50/50 chance, and in February this year SpaceX switched its focus to the Moon first, with Mars work starting in five to seven years. Its Mars page now says cargo flights start no earlier than 2028. So I was wrong about the timing. The Moon deserves an article of its own.
Meanwhile the “Pez dispenser”, the slot Starship uses to push out Starlink satellites, has already been tried out on test flights, and SpaceX plans to use it to put its first working Starlink V3 satellites into orbit. The same door could one day push out a rack of robots.
I’ve been wrong about dates before. But if you asked me today what to put in the hold, I’d still say leave the diggers at home and send the robots and the shovels. I’d love to hear from anyone who thinks I’ve got that wrong!