The Bold Gamble At The Core Of SpaceX's $1.75 Trillion IPO

SpaceX is pitching AI1, an orbital AI data center architecture, as a potential driver behind its $1.75 trillion IPO valuation, with each satellite designed to deliver 120 kW of continuous compute power. The company says Starship could deploy 30 to 50 satellites per launch and ultimately scale to 100 gigawatts of orbital AI compute annually, though key risks remain around radiation, thermal management, and launch economics. Prototype satellite launches are targeted for early 2027, making this a long-dated but potentially valuation-relevant strategic development.

Analysis

The core market takeaway is not that orbital compute is imminent; it is that a credible adjacency to AI infrastructure can justify an extreme multiple only if investors are willing to underwrite a long-duration option on supply-constrained compute. That shifts the valuation debate away from launch volumes and toward whether SpaceX can become a critical supplier of scarce inference capacity, which would make the story more comparable to a platform/cloud asset than a pure aerospace operator. The second-order beneficiary is NVIDIA: even a small probability of orbit-bound racks extends the addressable market for high-end accelerators and reinforces the premium attached to any GPU bottleneck.

The bigger competitive signal is actually negative for terrestrial incumbents that are still power-constrained: cloud operators and colocation providers facing multi-year grid interconnect queues may be forced into more aggressive capex, onsite generation, and long-term power contracts just to defend growth. That favors power equipment, gas turbine, and grid-services suppliers over traditional data-center landlords, while pressuring any name whose growth assumptions depend on cheap, quickly provisioned land + power. Microsoft is the cautionary read-through: the market can applaud the engineering but still discount the economics if upgradeability, serviceability, and customer adoption remain weak.

The key catalyst path is binary over the next 12–24 months: prototype success in 2027 is not enough by itself; investors need proof that thermal management, radiation tolerance, and launch cadence all converge into a repeatable unit cost curve. If launch frequency or satellite manufacturing slips, the whole narrative reverts to science project optionality and the valuation premium compresses quickly. Conversely, if SpaceX demonstrates even a modestly scalable orbital compute stack, it strengthens the case for a structural re-rating in adjacent AI infrastructure names and for capital to rotate toward enabling picks-and-shovels rather than pure software multiple expansion.

Key Decisions for Investors

• Maintain an overweight in NVDA versus broad AI/software baskets for the next 6-12 months; the orbital-compute narrative increases the terminal scarcity value of frontier GPUs, but size it as a momentum trade with a tight stop if AI capex growth decelerates.
• Initiate a tactical long GS versus MSFT pair for 3-9 months: GS benefits from IPO/financing activity and event-driven advisory flow if SpaceX-related capital markets enthusiasm expands, while MSFT is more exposed to investor skepticism around alternative compute architectures.
• Favor RDW/RKLB on weakness only as a speculative basket, not a conviction long; the thesis is a sympathy rerating on broader space-capex enthusiasm, but the probability-weighted upside is capped unless they show direct participation in orbital infrastructure supply chains.
• Use a long power-infrastructure / short data-center REIT pair over 6-12 months if interconnect queues stay tight: long utilities/grid equipment or gas-fired power enablers, short names most exposed to stranded data-center buildouts and delayed commissioning.
• For event-driven exposure, buy long-dated out-of-the-money calls on NVDA or a basket of AI infrastructure names into the 2027 prototype window; risk/reward is attractive because the market can reprice the option value of orbital compute well before commercial viability is established.