Scottish Mortgage's stake in SpaceX has dropped to just over 15% of the trust and more of it will come off the table in the weeks ahead, manager Tom Slater said today at the firm’s investor conference, as the trust looks to recycle the money into new ideas.
Scottish Mortgage first invested in SpaceX in 2018, when Elon Musk’s rocket company was valued at around $30bn. Its recent initial public offering (IPO) valued it at close to $2trn, and the trust's original $200m stake had grown to roughly $5bn, making it one of the largest single contributors to returns.
That is why the position is now well above a conventional level. Yet, Slater said investors "shouldn't be surprised by that”.
“Returns are driven by a small number of big winners, and our job is to try and maximise the return from those big winners for our shareholders,” he said. “That means positions will get large. We don't equate large positions with risk. It's more about can you run your big winners and make sure you capture all of that return for shareholders."
The investment case from here rests on falling launch costs, according to co-manager Lawrence Burns. Before SpaceX, putting a kilogram of mass into orbit cost $18,000 to $19,000. With Falcon 9 today, the list price is about $2,700 to get a kilogram into orbit and with Starship, it’s getting to about $900, he explained. “As they move to it being reusable, the ambition is to get that between $100 to $200 per kilogram.”
Burns pointed to Starlink as proof of what cheaper launches unlock commercially. The satellite broadband business is now generating revenue at a $17bn annualised run rate, growing nearly 70% year on year with operating margins approaching 40%.
"As that launch cost continues to fall to the hundreds of dollars, you start to unlock multiple Starlinks. You move into having compute and data centres in orbit. Defence applications become numerous. When you take a step back of what SpaceX is, it is, in effect, a near monopoly on the access to the rest of the universe."
Scottish Mortgage has begun selling SpaceX shares at the approved tranches, but this is not a lack of conviction in the stock but part of a broader portfolio reshuffle.
The trust exited Tesla entirely after 13 years, a holding Slater says returned around $6bn of profit and a 13-times multiple on invested capital, on the view that the valuation had come to reflect too much future opportunity already. Netflix was also sold out of, as growth matured and conviction faded.
Where the proceeds are being reinvested
New capital available through SpaceX share sales has gone into a mix of AI infrastructure and newer additions: more was added to Nvidia in chips, Cloudflare in internet infrastructure, Nubank in digital banking and Xie in e-commerce and digital finance.
New holdings in Vistra Energy and EQT Corporation were added for the separate – but related – energy theme, as AI demand grows the need for reliable power and data centre buildout is running up against how quickly new electricity supply can be built.
The rationale behind these additions is the idea that the more durable winners in AI will be at the chip level, not at the applications level.
“All these companies own bottlenecks within the supply chain. They're in dominant positions and able to develop really significant profitability and growth. Whatever applications become super successful in the next 10 years, whether that's autonomous cars, robotics, or coding agents, these companies benefit,” Burns said.
“The one thing every AI company we talk to comes back to is that we're going to need a lot more chips."
On the scale of capital pouring into AI infrastructure, Slater pointed to recent data from within the portfolio itself: SpaceX's x-AI-linked compute business, built out through data centre capacity, has signed deals with both Google and Anthropic.
Slater said the estimated return on that capital looked to be running at "two or three times the capital investment as an annual return on an annual revenue to that infrastructure" – a sign, he said, of just how strong current returns on AI hardware are and why the spending shows little sign of slowing.