Every gold rush produces two kinds of millionaires. The ones who found gold, and the ones who sold the pans. Elon Musk’s orbital AI bet is the most ambitious mining claim anyone has staked in years, and the company printing money off it doesn’t have to leave the ground.
That’s the setup investors are staring at right now. Analysts say Nvidia has more upside than SpaceX after Musk’s latest announcement. Nvidia shares have jumped nearly 10% in the last month. SpaceX stock has dropped more than 13% as excitement around its long-anticipated IPO cools off. Barron’s went as far as calling Nvidia the big winner from SpaceX’s earnings, which is a strange sentence until you think about who’s actually invoicing whom.
Reviewing hype is my job, and this pattern is familiar
I spend my days testing AI tools that promise the moon and deliver a Chrome extension. So I’ve developed a reflex: when a company announces something spectacular, I stop looking at the announcement and start looking at the supply chain underneath it. Who gets paid on day one, regardless of whether the vision works?
In AI agents, that’s usually the model provider. The startup wrapping a chat interface around someone else’s API lives or dies on retention numbers. The company selling the API gets paid either way. Compute is the toll booth, and toll booths don’t care about your business plan.
Orbital AI is the same structure at a much larger scale, with rockets. SpaceX has to launch the hardware, keep it functional in an environment that destroys electronics, handle thermal management without air, and build a business model on top of all of it. Nvidia has to ship chips. Guess which one has fewer failure modes.
The $28.5 trillion number is the tell
SpaceX’s filing reportedly lays out how the company gets to a $28.5 trillion revenue opportunity. That’s a total addressable market figure, and TAM figures are the oldest trick in the pitch deck. I see them constantly in AI startup materials: some vast number representing every dollar theoretically spendable in an adjacent category, presented as though it’s a forecast.
It isn’t a forecast. It’s a ceiling calculated under generous assumptions, and the distance between a ceiling and a floor is where investors lose money. The market seems to agree, given the 13% slide while the IPO approached its expected June launch date. Enthusiasm that cools before the listing is a signal worth reading.
Nvidia’s story is duller and that’s the point. The bull case is: sovereign AI programs, enterprise AI deployments, and AI-native startups all buying chips. Three demand channels, all of which already exist, all of which are already sending purchase orders. No orbital thermal engineering required.
What I’d actually be skeptical about
To be fair to the ambition, I don’t think Musk’s orbital bet is nonsense. Putting compute where the power is free and the cooling problem is different is a legitimate idea. Big infrastructure ideas that sound absurd have worked before, and SpaceX in particular has a habit of making absurd things routine.
But being right about a technology and being right about a stock are separate skills. Plenty of people were correct that AI would matter and still lost money picking which company would capture the value. The pattern I keep seeing in this space:
- The visionary layer takes on the hardest execution risk and the longest payback period.
- The supplier layer collects revenue from every participant, including the ones who fail.
- Public markets price the supplier’s certainty higher than the visionary’s optionality, especially when rates and patience are limited.
Nvidia’s other advantage is optionality without exposure. If orbital compute works, Nvidia sells the chips. If it doesn’t, terrestrial data centers still need the same chips. It’s a hedged position that requires no hedging.
My read as a reviewer, not a financial advisor
I review AI tools. I’m not qualified to tell you what to do with your portfolio and I won’t pretend otherwise. What I can tell you is that the analyst consensus here matches the pattern I see over and over in the AI tooling market, where the companies with the widest ambitions are almost never the ones with the healthiest near-term economics.
SpaceX is asking investors to buy a future. Nvidia is asking them to buy a present that happens to be attached to every plausible future, including SpaceX’s. When the visionary and the supplier are both on the table, the supplier tends to be the less exciting purchase and the more defensible one.
The orbital bet may well pay off eventually. In the meantime, someone has to ship the silicon, and that company already knows exactly who its customers are.
🕒 Published: