Last week's $SPCX Q2 earnings had one number that caught everyone's attention, the massive AI CapEx increase. I just published a full deep dive on why I think the AI segment could turn profitable much sooner than expected. Here are the key points… SpaceX is going all-in on AI, exactly like it went all-in on connectivity all of the sudden during Covid (and on reusability before that). Both bets worked. Connectivity is still the only profitable segment, with $1.66B of operating income this quarter, up 79% year-over-year. And why AI is comparable to Starlink is because AI in space will be as cheap to run as any connectivity business like Starlink, $IRDM, etc. Starmind will be $SPCX AI satellite that runs compute in orbit on solar energy. Once in orbit, no water cooling and no maintenance, unlike on Earth. The main costs are manufacturing, deployment and ground infrastructure, exactly like the profitable Starlink segment. Deployment will be super cheap because SpaceX has the advantage of having the world’s first fully reusable and most powerful rocket ever built with Starship. So just like planes, fuel and maintenance is the main cost. Connectivity segment tends to be a passive income generator for companies (once in operation). $IRDM that is going to be acquired for $8 billion by $RKLB mid-next year has 57% of OEBITDA margin. To highlight some of the risks of the early stage of a connectivity business, $IRDM previously went bankrupt in 1999 after its massive $5 billion satellite network could not attract enough paying customers to service its heavy debt. The AI demand in upcoming years will be key for Starmind to succeed. AI revenue hit $2.6B, up 213% from Q1 and 247% year-over-year. Most of it comes from three cloud contracts: Anthropic at $1.25B/month, $GOOGL at $1B/month and Reflection AI at $150M/month. That's $2.32B monthly, or $27.8B annualized, from only three deals. The AI segment is now EBITDA positive and the loss from operations went down year over year even with R&D doubling. Once Starmind is operational and R&D comes back down, operating income for the AI segment is not too far away. To build the Starmind satellites at scale, SpaceX is building a $55B Gigasat factory in Bastrop, TX, with thousands of satellites to be deployed as soon as late 2027 and Terafab in Grimes County, TX for the chips ($16.8B first phase, 3,000 jobs), which they've broke ground last week. The first version of Starmind will run only on $NVDA Rubin GPUs and Vera CPUs, since Musk said they are now exclusive to Nvidia. On the launch side, Starship 13th flight test deployed last month the first Starlink V3 satellites (in a suborbital trajectory) and did a very soft splashdown in the Indian Ocean, so the team could inspect the ship closely for the first time. Starlink V3, to begin being deployed at scale later this year, should bring about 100x more bandwidth. Management said that even if revenue per bit would hypothetically fall by 90%, they could still make 10x more revenue thanks to the extra bandwidth. So the decreasing ARPU from $85 to $66 per subscriber does not seem like a concerning factor, especially since subscribers have doubled from 6 to 12 millions on a year-over-year period and per-bit monetization would increase massively once V3 satellites are deployed. On valuation, at $1.75T and roughly $73B annualized revenue, $SPCX trades around 24x sales. Expensive, but looking at other companies like $ASTS the PS is at 135x sales for about $200M of revenue this year. So SpaceX actually looks less priced-in than a lot of other "high growth" more speculative companies in the area. Of course, they are very different companies. SpaceX does not really have a direct competitor when it comes to AI in orbit, so it is difficult to compare it directly with another company. $ASTS is still in pre-revenue while $SPCX has been generating billions for years now. Musk pulled the $1T revenue target forward to 2030 and is guiding a $100B run rate by year-end by doing "nothing" more differently. That means going from about $18B to $25B in quarterly revenue in two quarters. Looking at how fast the AI segment is growing (+213% increase from last quarter), and if more compute contracts get signed by other AI companies or even hyperscalers, I think the $100B run rate by end of year is highly very likely, but of course the optimistic $1T revenue target will depend on the deployment timeline of the Starmind satellites in coming years, too early to say. Argus analyst Steven Silver upgraded the stock to “buy” from “hold” with a $160 price target and expects that the rapid monetization from these massive cloud contracts will allow revenue to outpace and eventually fund the AI infrastructure. Am I buying $SPCX? Not until I see it at $700B valuation, which may happen with time or not, but I have it through $ORBX. *not a financial advice* Read my full article https://bdinvesting.substack.com/p/spacex-is-closer-to-profitability @bdinvesting
Let me start by saying Starlink is an amazing 'business unit' within $SPCX and would be a terrific standalone business and for that matter stock IMO... As space AI datacenters go, I personally don't think Starmind's deployment & operation costs will be nearly as cost effective as Starlink's been. I might not be cost effective at all IMO. TBD I guess... Datacenter size alone makes me skeptic, let alone all the technological challenges like thermal management, hardware degradation due to radiation, latency and costs (the latter of which I know you don't agree with). Time will tell how this all plays out. I enjoy hearing/reading differing views and I hope you don't mind me sharing my opposing view here. It's obvious you always put a lot of effort into your posts and I appreciate that, whether I agree or not. Cheers Anthony!
Mike L@noviceadvisor · 2h
But you said you wouldn't touch it with a 10 foot pole previously 😆
Join the conversation with 500,000+ other investors 💸
Create an account to get access to everything Blossom has to offer!
📊 Personalized algorithm based on your investing style, experience level and interests
📈 Powerful portfolio and dividend tracking tools
👀 See what top creators and others in the community are investing in
