Three Prices for the Orbital Data Center Bet

When the Danish pension fund AkademikerPension blacklisted SpaceX's IPO in May, it took care to say what the decision was not about.

Our decision not to invest is therefore not a reflection of the quality of its technology or engineering expertise. — Magistrenes Pensionskasse

The $25 billion fund was walking away from a governance structure in which Elon Musk's Class B shares carry ten votes each, granting him 85.1% voting control and an effective veto over his own removal as CEO and chairman ^[1]. A coalition of U.S. institutional investors — New York City Comptroller Brad Lander, New York State Comptroller Thomas DiNapoli, and CalPERS CEO Marcie Frost — soon echoed the concern ^[2].

The current proposed structure makes it nearly impossible to ensure strong safeguards are in place to preserve the company’s financial and reputational value, limits transparency, thwarts the opportunity for accountability, and overall, dangerously undermines investor rights. — Mark Levine

The technology, AkademikerPension was saying, was not the problem. And the last two months have borne that out — in prices. The same unproven concept — solar-powered orbital AI data centers, a constellation of satellites running compute workloads above the atmosphere to bypass terrestrial electricity and water constraints — is being funded at three valuations with a 7,000-fold spread between the top and the bottom. SpaceX priced it at $2 trillion in its June IPO filing, which named "up to one million space-based data center satellites" as a primary use of proceeds ^[3]. Blue Origin is raising its first external capital — $4 billion from Coatue Management, $2 billion from Jeff Bezos — at a $130 billion valuation for Project Sunrise, its own proposed constellation of 51,600 orbital AI satellites ^[4]. And Cowboy Space Corporation raised $275 million in Series B funding led by Index Ventures for the same thesis ^[5]. The technology is uniformly speculative. Launch costs remain above the $200-per-kilogram threshold that industry experts say is needed for economic viability, and no one has deployed a working orbital GPU cluster ^[5]. The difference is not the readiness of the technology. It is the terms attached to each vehicle. SpaceX's IPO filing disclosed a dual-class share structure that corporate governance experts said exceeds typical norms for founder-led technology companies ^[1]. The company was losing $4.3 billion per quarter, driven by Starship capital expenditures and the absorption of the unprofitable xAI business ^[6]. In late June, it announced a $20 billion senior unsecured note offering and a $60 billion all-stock acquisition of the AI coding startup Cursor — shares fell 16.4% in a single session, erasing roughly $600 billion in market value ^[7]. S&P Global assigned a BBB rating and projected SpaceX would remain cash-flow negative until 2030 ^[7]. Then there was the revenue forecast. Musk projected SpaceX would reach $1 trillion in revenue by 2031 ^[8]. The company's own IPO underwriters disagreed sharply: Goldman Sachs forecast $470 to $474 billion for 2030, Morgan Stanley projected $330 billion, and Ark and New Street Research ranged from $195 to $400 billion ^[8]. Even the most bullish sell-side estimate came in at less than half of Musk's target. The divergence between operational achievement and market judgment reached its clearest point this past week. On July 24, Starship Flight 13 succeeded — deploying 20 Starlink V3 satellites and achieving a soft ocean splashdown ^[9]. Three days later, SpaceX shares hit an all-time low of $108.66, down nearly 50% from the June 12 IPO peak of $225.64, erasing over $1.2 trillion in market value ^[9]. The decline was accelerated by a broader four-day tech selloff that erased $1.2 trillion from the sector, driven by AI spending concerns and a Jefferies warning of "massive capital destruction" as Chinese open-source models outperformed U.S. counterparts ^[10][11]. But SpaceX's 50% drop was far steeper than the 5% to 16% declines across other tech stocks — company-specific factors were the primary driver ^[10]. Musk posted on X.

(Former) Trillionaire — Elon Musk

There is genuine skepticism about the orbital data center concept itself. SoftBank CEO Masayoshi Son rejected it in June, arguing that electricity savings from space-based solar are offset by prohibitive hardware, transportation, and maintenance costs, plus communication latency ^[12].

In the battle for AI, the next few years will be far more important than what might happen a decade or so from now. — Masayoshi Son

OpenAI CEO Sam Altman dismissed the idea as "ridiculous" in February, citing high launch costs and the impossibility of repairing GPUs in orbit ^[13]. These are technology objections, and they are serious. But they did not stop Coatue from writing a $4 billion check to Blue Origin two weeks after Son's remarks, or Index Ventures from leading Cowboy Space's $275 million round. The technology risk is real — and a set of investors is taking it, just at valuations that do not carry SpaceX's governance and financial baggage. The market is not speaking with one voice. Ron Baron's Baron Capital held a $1.7 billion private stake in SpaceX, added $1 billion at the IPO, and predicted the company could be worth ten to thirty times its post-IPO value over the next decade ^[14]. Cathie Wood's Ark Invest sold Alibaba and AMD positions to buy SpaceX after the debut, then added $80 million more during the pullback ^[15][16]. Raymond James issued a strong buy with an $800 price target ^[16]. These are not marginal players, and SpaceX even at $108 a share still trades at a massive premium to Blue Origin's $130 billion. But the investors who declined — AkademikerPension, the U.S. institutional coalition — did so on governance grounds, and the investors who wanted the orbital data center bet without the SpaceX premium found it elsewhere. Blue Origin's $130 billion round and Cowboy Space's $275 million raise are the same concept, priced without an 85% voting concentration, without a veto over the CEO's removal, without $4.3 billion in quarterly losses and $20 billion in fresh debt. The last two months have sorted out what price the orbital data center idea can command under cleaner terms. The idea is drawing capital. The terms are the spread.