SpaceX’s latest FCC filing signals a bold escalation in the battle for AI infrastructure dominance: a proposal to deploy up to 1 million solar‑powered orbital data‑center satellites — positioning space, not Earth, as the next frontier for compute capacity. Framed as “the most efficient way to meet the accelerating demand for AI computing power” and even “a first step towards a Kardashev II-level civilization,” the plan underscores a massive shift in how hyperscale workloads could be powered and scaled, bypassing terrestrial constraints such as electricity, cooling, and land availability. But this ambition collides with some harsh realities and questions the timing of this filing — arriving alongside other company news.

Here’s our take on a few key topics and what to expect next:

  • Classic Musk valuation inflation. Layered beneath the technical bravado is a clear capital‑markets tell: By aligning an FCC ask for 1 million orbital AI satellites with news of a SpaceX-xAI combination ahead of a potential IPO, SpaceX is packaging a narrative of hyperscalable, lowest‑cost AI compute from space — the kind of storyline that expands TAM, lifts growth multiples, and reframes Starship as a structural cost moat. These are classic ingredients for pre‑IPO narrative lift — and a familiar pattern in Musk‑era market signaling. It positions ambition as inevitability, inviting investors to price in a future that engineering has yet to substantiate.
  • The price tag is massive — and the timeline is even harder to swallow. The arithmetic is sobering. Even using Starlink‑era proxies, hardware alone ranges from $250 billion to $1 trillion before launch and refresh cycles. At today’s demonstrated launch pace — and even granting Starship lift — deploying 1,000,000 units would take five decades at ~145 flights/year and ~125 sats/flight, implying an order‑of‑magnitude scale‑up in pads, range ops, manufacturing, and regulatory throughput to compress into a decade. Visionary, yes — but until flight rate, unit cost, and regulatory capacity bend simultaneously, tech leaders should treat this as aspirational rather than executable.
  • A crowded orbit and constant replacements. What could go wrong? This mega‑constellation amplifies long‑standing concerns about Kessler syndrome, the cascading orbital‑debris chain reaction that could render key orbital bands unusable for generations. Such fears have already resurfaced amid recent satellite fragmentation events and ballooning congestion in low Earth orbit — a region where space junk is already a worsening threat. These satellites would also face the same short operational lifetimes and continual replacement cycles seen in today’s Starlink fleet, which routinely deorbits aging units — further compounding congestion and debris‑management pressures.
  • The real bottleneck could be heat. SpaceX’s vision runs headfirst into a physics bottleneck: Radiative cooling in a vacuum demands enormous surface area, especially at multimegawatt power levels, because space offers no convection or conduction to shed heat. Credible engineering references confirm that radiator size scales sharply with heat load, making megawatt‑class thermal rejection systems vastly larger than anything on today’s satellites. For any constellation of high‑power AI nodes, that thermal burden would balloon mass, expand collision cross‑section, and turn heat management — not launch capacity — into the gating constraint for orbital data centers.
  • Orbiting AI, earthbound power — in one man’s hands. Beyond debris risks, the prospect of AI compute shifting off planet raises thorny data sovereignty and governance questions: Who controls data processed outside national jurisdictions, how can regulatory bodies enforce compliance in a multiorbit ecosystem, and what safeguards can ensure accountability when computation occurs beyond terrestrial legal boundaries? The SpaceX proposal concentrates immense control in one company, turning a CEO’s decisions into geopolitical leverage. With China and others likely to launch rival constellations, interference and strategic tension will only intensify. Without new governance, it risks shifting from a shared AI domain to a geopolitical flashpoint.

Clients can engage with us via an inquiry call by emailing inquiry@forrester.com and read Forrester’s The Future Of Satellite Connectivity report.