The Mars to Moon Pivot
SpaceX has changed direction - were the haters right?
It has been some time coming, for those paying attention, but Elon Musk has announced that SpaceX is now prioritising the Moon over Mars, at least for the short term. The plan is to focus on lunar development and space based data centres, and then start building up the Mars base in 5 years (i.e. the 2031 launch window).
To many, this is vindication of their belief that Mars was always a “scam”, that anybody who believed it was happening was stupid, and that they were clever enough to see The Truth. Its a conspiratorial mindset, and unsurprisingly people holding it will consider almost anything to validate their belief that the entire SpaceX enterprise is a house of cards ready to collapse at any moment.
The reality, of course, is a little different.
The New Mars Timeline
Its clearly stated in this post that large scale Mars missions will begin in the 2031 or 2033 windows - which is what most people paying attention had already assumed. The setbacks of last year had made the 2026 window essentially unattainable, and the mission has always called for sending an unmanned flotilla in the synod before the human mission (that would be the 2029 window). I suspect it would take two synods to build full confidence to send humans - the first would be a data gathering mission, to test that the vehicles really can land safely, and the second would be vehicles refined based on this data delivering large amounts of cargo for the initial human missions. So I thought it would probably be 2033 for people, but Musk still publicly aims for 2031:
Slips in the development in Starship have pushed the Mars mission back many times - the initial pitch, back when it was a very different vehicle made out of carbon fibre, mentioned initial cargo missions in 2022. That the best case for these flights is now 2029 is often seized upon as proof that the entire proposal is a fraud meant to boost the stock price (of a company that is not yet publicly traded) by promising to spend perhaps a trillion dollars in the future with little chance of return. These critics don’t seem to understand how business and investment works very well.
The more mundane explanation is of course that Starship is a big, complex project and has suffered delays and setbacks, as these things often do. It would be naive to assume that delays will continue to grow at this stage, as many of the problems of the program are now behind it. Citing past delays to claim the program will keep getting delayed indefinitely, without any technical insight into the actual progress made, is silly.
The AI Data Centre Rush
The reason Falcon 9 has been able to achieve such high flight rates, high levels of reuse, and amazing economics is Starlink. There was not enough external demand in the launch industry to drive the cadence ramp SpaceX wanted, so they generated demand internally. Not only did Starlink turn out to be very profitable in itself, it also helped reduce costs of launch itself.
The possibilities of space based data centres offer a chance to do this again for Starship, at a much larger scale, hence the company absorbing xAI. The concept is that there will be huge ongoing demand for compute in space, so much so that it will not even be possible to meet it through Earth launched equipment, and a Lunar base with a mass driver will be needed to continue the exponential increase. If it pans out, this will provide a commercial rationale for thousands of Starship launches, paying for the acceleration of the program, and this high cadence vehicle will then be available for Mars missions when a window arrives.
Now the viability of this assumption is up for debate - not so much due to the possibility that we are in an AI stock bubble, because that just means particular companies are overvalued and prices will correct, but because there are differing views on whether or not current AI approaches will slam into some kind of hard barrier causing advances in intelligence to slow. The vibe I get from experts on this is that the answer is “no”, but your mileage may vary. Musk is betting the company somewhat on continued rapid advances in the field requiring ever more compute to train meaningfully more capable models.
If this bet pays off, and although I am no expert in AI I think it has a better chance than not, then it means the demand for Starship flights is practically unlimited and SpaceX can profitably launch them as fast as it is physically capable of doing so. And that same armada of reusable rockets will feed directly into the Mars effort.
Solving the Mass Problem
Elon Musk understands, as many do and as I myself have written about here, that the fundamental problem in space is payload mass. Almost everything is downstream of it. You want to go further or get their faster? You need more propellant mass, and more massive tanks to store it, and more massive engines to provide sufficient thrust. You want to keep astronauts alive for long duration space voyages? They require a lot of mass in terms of food, water and oxygen. Radiation a problem? You need more mass for shielding.
Solving the mass problem is the necessary but not sufficient condition for solving all problems in spaceflight, and in pretty much every case there is a sufficient solution that involves throwing more mass at the problem. This mass doesn’t have to come from Earth, although for the moment it almost always does. It can be provided by recycling waste, as is done on the ISS to some extent, or from in space resources.
The critical thing here is that a good solution to the mass problem (such as a high Starship cadence, or a lunar mass driver) is applicable to pretty much every mission you could want. It will help satellite companies, habitat builders - and Mars colonists. So long as SpaceX remains focused on this problem, they will continue to be successful.
Thank you for your calm and well-reasoned posts.