A billboard organized by corporate accountability group Eko passes through Westminster urging the Prime Minister to stand up to Elon Musk and ban X and Grok, in London, England, on Jan. 14.Maja Smiejkowska/Reuters
Michael Byers co-directs the Outer Space Institute
One of Elon Musk’s companies, SpaceX, recently acquired another of his companies, xAI.
SpaceX is now the world’s largest privately held company, worth an estimated US$1.25-trillion. An IPO, expected later this year, could raise US$50-billion in new equity capital.
It’s all very exciting, but investors might wish to be cautious.
As the name suggestions, xAI is focused on artificial intelligence. The company has spent heavily on data centres as it races to catch up with OpenAI, Anthropic and other industry leaders. It has been underperforming, ranking behind competitors on industry benchmarks. On top of that, its chatbot, Grok, has been caught up in controversies over it generating hate speech and sexualized images.
SpaceX also has problems. Its revolutionary technologies – reusable rockets and a “mega-constellation” of nearly 10,000 broadband satellites – are being replicated by competitors with their own deep pockets.
Elon Musk’s xAI overhauls top management, lays off employees
Blue Origin, owned by Jeff Bezos, recently launched and relanded an orbital rocket. Amazon, also owned by Mr. Bezos, has started to launch its first mega-constellation and is preparing for a second.
Chinese companies such as LandSpace and Shanghai Spacesail Technologies Co. are also catching up, testing a reusable rocket and launching two mega-constellations.
The global market for space-based broadband will be finite, and militaries are the largest customers. Outside of the United States, governments have stopped trusting Mr. Musk after he limited coverage from his Starlink mega-constellation over the Black Sea, deliberately constraining Ukraine’s ability to attack Russian warships.
As soon as other mega-constellations are available, governments in Europe and elsewhere might well take their business elsewhere.
SpaceX has also spent vast amounts of money developing Starship, but the world’s largest rocket keeps exploding during testing. Initially scheduled to land on the Moon in 2024, Starship has fallen years behind schedule, causing NASA to consider other options.
A successful IPO would throw a lifeline to Mr. Musk, and this is where a struggling AI company and a delayed rocket come together: in a plan to use Starship to launch a million data-centre satellites to power AI.
According to SpaceX, “By directly harnessing near-constant solar power with little operating or maintenance costs, these satellites will achieve transformative cost and energy efficiency while significantly reducing the environmental impact associated with terrestrial data centers.”
U.S. Senators ask Pentagon to probe SpaceX for potential Chinese ownership
Now, it is true that data centres consume vast amounts of electricity, and solar-powered satellites can harness the sun without interruption, if placed in orbits that keep them out of earth’s shadow. It is also true that data centres require copious amounts of water for cooling, unless located in an already cold place such as the Arctic or space.
Yet Mr. Musk’s plan has serious weaknesses. SpaceX will not be able to access the space-based data centres to maintain or upgrade them. They will therefore have short operational lives and require constant replacement.
Mr. Musk has already taken this “consumer electronic product” approach to the Starlink mega-constellation: Its satellites last just five years, with replacements requiring a constant cadence of rocket launches.
What’s more is that Mr. Musk’s plans will result in a loss of safe access to space for everyone, with catastrophic consequences for economies that depend on satellites for everything from navigation to crop monitoring to precision timing for financial transactions.
SpaceX has nearly 10,000 Starlink satellites in orbit and plans to launch 20,000 more. All these satellites interfere with ground-based telescopes by reflecting sunlight into the shadowed, nighttime side of the Earth. Astronomy, which includes the detection of asteroids on Earth-impact trajectories, is becoming increasingly difficult to practise.
Meanwhile, space debris results from satellites and rockets being abandoned in orbit, and from collisions between these objects. More than a million pieces are already present, most of them too small to be tracked with radar, and all this debris crosses the orbits of operational satellites at a menacing seven kilometres a second.
The risk of a collisional cascade is growing, as each collision creates more debris, leading to more collisions, and so on. And now, Mr. Musk’s plan for a million space-based data centres would increase the number of satellites in orbit by a factor of 50.
Fortunately, a more sustainable approach is possible. Instead of replicating Starlink, new broadband constellations could be built out of smaller numbers of higher-quality, higher-capacity and longer-lived satellites. This is the approach taken by Ottawa-based Telesat, with a soon-to-launch constellation of just 156 satellites.
As for data centres in space, the monetary costs of launching and then constantly replacing them will be monumental, while externalities such as light pollution and space debris will, sooner or later, lead to regulatory constraints.
Earth-based solutions – solar farms in deserts, data centres in the Arctic, high-voltage direct current transmission lines and fibre-optic cables – may offer a safer, more financially secure approach.