SpaceX went public at $1.75 trillion. Now three stocks get repriced.
The problem is that most institutions that want space exposure now face a familiar constraint. SpaceX is priced for perfection at launch. The obvious next move for capital managers who missed the IPO window, or who want a second bite at the sector, is to rotate into the handful of public companies that orbit (figuratively) around SpaceX's dominance.
Rocket Lab USA (NASDAQ:RKLB) is the most direct play. It is the only other company on the planet currently flying an operational orbital rocket on a commercial basis. When SpaceX was private, RKLB was a speculative bet on whether a second launch provider could ever build a sustainable business. Now that SpaceX has a public market cap to anchor against, RKLB has a valuation reference point. Institutions running space thematic funds need more than one name. RKLB is the only credible candidate.
AST SpaceMobile (NASDAQ:ASTS) is a different angle. Its thesis is satellite-direct-to-phone broadband, which is a very close cousin to what Starlink proved could be built and monetised at scale. The difference is that AST targets standard smartphones, not proprietary terminals. A richly valued, publicly traded SpaceX validates the satellite connectivity market in a way that no analyst report ever could. Capital that believes in Starlink's commercial model but finds the entry price uncomfortable will look hard at ASTS.
Intuitive Machines (NASDAQ:LUNR) is the most unusual of the three because it is, in a sense, a SpaceX customer. Its lunar landers fly on Falcon 9 rockets. Its NASA contracts, worth billions, depend on a launch provider that can actually get to the Moon reliably. LUNR's fortunes and SpaceX's are operationally linked, and now that SpaceX trades publicly, the market has a cleaner way to price what that relationship is worth.
The chain reaction here matters. Pre-IPO, space was priced as speculative. Post-IPO, at $1.75 trillion, it is priced as infrastructure. That reclassification changes how pension funds, sovereign wealth vehicles, and large family offices model the sector. They do not chase SpaceX after a $1.75 trillion listing. They build positions in the companies that grow inside SpaceX's gravity.
All three of these stocks carry real risk. RKLB is burning cash to scale its Neutron rocket. ASTS still needs to prove commercial viability at the network level. LUNR is dependent on NASA contract continuity in a budget environment that has its own pressures. These are not safe bets. But if the SpaceX IPO truly marks the moment that institutional capital treats space as a permanent asset class, the repricing of these three names has probably only just begun.
Three more stories worth knowing about
MU's 34-point score drop in four days is worth watching
Anthropic's fall IPO will be the AI sector's next re-rate moment
Golden Dome defence spending is already inside the budget lines
Celestica, Inc.
Celestia is a contract manufacturer and supply chain solutions company headquartered in Toronto, serving hyperscalers, aerospace and defence clients, and the AI infrastructure build-out. Its collaboration with AMD on Helios, a rack-scale AI platform, puts it directly inside the hardware layer that data centres are scrambling to deploy.
The business model is unglamorous. Celestica does not design the chips or write the models. It builds, integrates, and ships the physical systems those chips live in. But margins in that space have been improving steadily as hyperscaler customers consolidate their supplier lists around trusted partners with proven manufacturing depth.
At a 94 score with a True North tier rating, this one warrants a closer look before the AI infrastructure theme gets crowded enough that everyone already knows the name.
NVDA's score drop is noise. The demand story has not changed.
Consensus is starting to murmur about demand softening, margin pressure as custom silicon from hyperscalers matures, and valuation stretch. Some of that is legitimate.
Here is what the bears keep getting wrong: the constraint on AI infrastructure spending is not demand. It is supply. Every major hyperscaler has publicly committed to capital expenditure on AI compute that dwarfs anything they have spent before. Microsoft, Google, Amazon, and Meta have all guided for it. They cannot deploy that capital without Nvidia's GPUs, because custom silicon takes years to reach the performance density Nvidia ships today.
A 16-point score dip over four days is trader noise. The 2027 order book has not moved. Unless something structural breaks in the HBM supply chain or a hyperscaler materially pulls guidance, this is the kind of dip that looks obvious in hindsight.