SpaceX and NVIDIA Want to Put AI Data Centers in Orbit

SpaceX and NVIDIA Want to Put AI Data Centers in Orbit Written by Sam Quirke | Reviewed by Clare Titus September 20, 2026 Add As Preferred Source Share Share Share This Article Link copied to clipboard. Close Key Points Elon Musk says he is highly confident SpaceX will begin launching NVIDIA's advanced AI computers into orbit as early as 2027. SpaceX would adapt Starlink satellites with NVIDIA chips and solar power, aiming for larger deployments by 2028 and eventually up to a million satellites. Significant obstacles like heat dissipation, radiation shielding, and repair logistics mean analysts do not expect meaningful revenue before around 2029. Five stocks we like better than SpaceX . Some corporate ambitions are so audacious they sound like science fiction, and Elon Musk's latest may be one of them. It was reported earlier this week that he's “highly confident” his rocket company, SpaceX NASDAQ: SPCX , will begin launching NVIDIA Corporation's NASDAQ: NVDA most advanced AI computers into orbit as early as next year. If it happens, this will be the first step toward a future where artificial intelligence is powered not by data centers on Earth, but by satellites circling above it. SpaceX Today SPCX SpaceX $152.71 -2.10 (-1.36%) As of 09/18/2026 04:00 PM Eastern 52-Week Range $104.83 ▼ $225.64 Price Target $221.42 Add to Watchlist The plan itself is not new, but Musk's confidence in a 2027 start is, and that near-term timing is what makes it worth a fresh look. It sounds outlandish, and in many ways it is. Yet it ties two of the market's most closely watched companies to a wholly new frontier for computing, and for two stocks that could each use a fresh catalyst, this is one to watch. Get SpaceX alerts: Sign Up Both have had a decent run of late, with SpaceX shares up around 45% since the start of August and NVIDIA climbing roughly 10% over the same stretch, though both are roughly flat over the past month. A bold new growth story could give each of them a reason to break out , which is precisely why this orbital plan, however far-fetched, deserves a closer look. What Musk Is Actually Proposing Strip away the theatrics, and Musk’s idea has a certain logic. AI data centers are voracious consumers of electricity, water, and land, and they're starting to hit real constraints on Earth, from strained national power grids to local planning battles. Space, his argument goes, offers a way around all of it: limitless room and solar power, and no neighbors to object. The plan is for SpaceX to adapt the satellites it already builds for its Starlink network, swapping their communications gear for NVIDIA's computing hardware and adding solar panels to power it. Musk has spoken of an initial launch late next year, with larger deployments following in 2028. He has also floated the staggering ambition of a constellation of up to a million compute-carrying satellites. Crucially, this is much more than just a SpaceX story. NVIDIA has been named the exclusive supplier of the chips that would think in orbit, so any buildout, however far off, would flow straight into fresh demand for its hardware. And with NVIDIA already holding a sizable stake in SpaceX, its interest runs deeper than a typical supplier's, giving its own shareholders a real reason to follow where this plan goes next. The Case for Getting Excited For the bulls, much of the appeal comes from what SpaceX already owns. It builds the rockets, makes the satellites, and runs the Starlink network, and all of that expensive infrastructure is already up and running. That means it could add a whole new business, orbital computing, on top of assets it has already paid for, at a marginal cost no rival starting from scratch could hope to match. The upside for NVIDIA is simpler, and arguably even bigger. Any large-scale orbital buildout would mean a vast new source of demand for its chips, on top of the already insatiable appetite from earthbound data centers. That's what makes Musk’s idea so tantalizing for investors: two companies at the cutting edge of their fields, potentially joining forces to open a market that doesn't yet exist. Why the Skeptics Are Right to Be Wary Before getting carried away, however, a heavy dose of realism is in order, because the obstacles are formidable. Chief among them is the simple question of heat. In the vacuum of space, there is no air to carry warmth away, so cooling powerful computers would be seriously difficult, with one estimate suggesting a single orbital data center might need radiators covering millions of square feet. Then there's everything else: shielding delicate hardware from radiation, moving huge volumes of data between orbit and the ground, the near-impossibility of sending a human engineer up to fix a broken machine, and the thorny questions of regulation and insurance for a business with no track record. None of these is trivial, and together they are daunting. Most important of all is timing. Even the more bullish analys