SpaceX's orbital AI bet is a long way from the public market's existing vote
Reuters reports SpaceX is targeting late-2027 orbital AI tests. Meanwhile, the prediction market for the listing has quietly given up a third of its premium in three weeks.
By 10 June 2026, the loudest SpaceX story on the wire was not a launch. It was a date. Reuters reported on 2026-06-10 at 07:50 UTC that SpaceX had set an internal target of putting orbital AI computing tests in space by the end of next year — late 2027 — citing people familiar with the plan. The same framing was carried hours earlier by prediction-market feeds and, overnight, by CoinDesk's coverage of a derivative that has spent the last three weeks repricing the company's eventual public listing.
Read the two stories together and a sharper picture emerges. SpaceX is no longer just a launch services company with a sideline in Starlink. It is asking investors — both retail on a permissionless exchange and, soon, the institutional kind — to underwrite a compute-in-orbit thesis. The bet is that the cheapest place to run frontier AI workloads is 500 kilometres up, where cooling is free, power is uninterrupted, and the latency tax of beaming results back to Earth is a problem solvable with a few more ground stations.
What Reuters says
The Reuters report is short on the engineering and long on the corporate choreography. SpaceX is reportedly aiming to launch orbital AI computing tests by the end of 2027, with the work folded into the broader Starlink and Starsheet operational stack. The plan, as described, treats orbit as data-centre real estate: a place where solar arrays run at near-constant duty cycle and the atmosphere that complicates terrestrial cooling and power is simply absent. Reuters's sources cautioned that the timeline is internal and could slip — a familiar caveat in a company that has historically under-promised on Starlink cadence and over-delivered relative to legacy aerospace.
The reporting is consistent with what the rest of the public record has been saying for months. Elon Musk has, in public settings, framed Starship as the missing piece for any meaningful orbital infrastructure; the orbital-AI timeline is best read as the first concrete customer for that vehicle, not as a separate programme.
What the market is doing in the meantime
The more interesting story is the one CoinDesk flagged the same morning. The SPCX perpetual contract on Hyperliquid — a synthetic that lets traders take positions on SpaceX's pre-IPO valuation without holding the private shares — has fallen 27% over three weeks. It still trades above the $135 price tag attached to the latest tender offer, but the premium has thinned dramatically from its May highs, when first-day-pop expectations were at their most aggressive.
That is a meaningful signal, and not the one a headline skim would suggest. Traders are not saying SpaceX is worth less than insiders said it was. They are saying the probability-weighted expectation of an immediate, explosive public listing has come down. A 27% drawdown in a perpetual that pays no yield and tracks sentiment more than cashflows is, in practice, a vote on timing.
The structural read
Two patterns are running in parallel. The first is the build-out of the orbital compute thesis itself — a long-duration capital project with no clear peer in the listed universe, and one that benefits from SpaceX's vertical integration in a way that no spin-out can replicate. If orbital AI works, SpaceX captures most of the value because SpaceX owns the launch, the bus, the spectrum, and the ground segment.
The second is the maturation of off-exchange pre-IPO trading. Hyperliquid's SPCX book is now a real-time sentiment gauge on a private company whose last private round reportedly valued it near $400 billion. Three weeks ago, that gauge was pricing in a near-term listing with a fat pop. Today, it is pricing in something closer to "sooner or later, and the pop will be smaller than the optimists thought." That is a healthier market in some respects — it is no longer treating a private valuation as a fait accompli. It is also a market that is, for the first time, allowing public traders to disagree with the company's chosen bankers on cadence.
Stakes and what to watch
If the Reuters timeline holds, the first orbital-AI test article is roughly eighteen months out. That is a long runway for a narrative to be repriced, and the Hyperliquid contract will move on every Starship test, every Starlink V3 milestone, and every leak from the company's internal cap table. The bull case is straightforward: SpaceX executes, the IPO prints at a multiple that vindicates the perpetual longs, and the orbital-AI story becomes the next leg of the thesis. The bear case is that the timeline slips, the perpetual continues to bleed premium, and the company ends up tapping the private market at flat or down rounds in 2027 — a possibility that becomes more painful the closer 2026 draws to its end without an S-1.
The honest read is that both can be partly right. SpaceX is genuinely closer to orbital compute than any listed competitor, and the public market's appetite for that story is, on present evidence, more cautious than the company's own.
Desk note: Monexus framed the Reuters report and the CoinDesk perpetual move as two halves of the same question — what does a private company's public-market shadow cost when the listing slips? Wire coverage tended to treat them as separate stories.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- http://reut.rs/4uVZjdf