Highlights

• Chamath Palihapitiya frames the SpaceX-xAI merger as a potential $1.25 trillion deal centered on orbital data centers to ease terrestrial power constraints.
• The ambitious 2027-2028 timeline faces significant technical hurdles including:
  • Radiation-hardened hardware
  • Thermal management
  • Unprecedented launch cadence requirements
• The vision's success hinges on unproven capabilities:
  • Achieving 100kW compute density per ton in space
  • Radiative cooling at scale
  • Near-hourly Starship launches
  • Robust orbital networking
• Each capability carries only a 20-60% probability of on-schedule execution within five years.
• A critical blind spot in the thesis is supply chain dependency:
  • Orbital infrastructure would dramatically increase demand for permanent magnets and rare earth materials.
  • These materials are still dominated by China.
  • Materials security is required, which the current U.S. industrial policy hasn't yet achieved despite recent Trump administration efforts.

---

In a recent Substack (opens in a new tab), Chamath Palihapitiya—a seasoned technology investor with a well-documented record across venture capital, public markets, and SPACs, and a significant shareholder in MP Materials—frames the SpaceX–xAI combination as the largest corporate merger ever contemplated, implying a valuation of roughly $1.25 trillion. His core thesis is ambitious and characteristically forward-looking: that “orbital data centers” could shift AI compute into space, easing terrestrial grid constraints through near-constant solar power and radiative cooling. It is a bold synthesis that reflects Palihapitiya’s strength in systems-level thinking—yet it rests on assumptions that deserve careful scrutiny.

Context—Bold, Aggressive

Elon Musk’s implied timeline for orbital AI data centers is bold, directional—and far from fixed. Based on public statements following the SpaceX–xAI merger, the iconic tech entrepreneur has suggested that space-based compute could become cost-competitive with terrestrial infrastructure within roughly 30–36 months, implying a 2027–2028 window for early operational viability, driven by near-continuous solar power and radiative cooling advantages. Independent analysts generally agree this timeframe represents the earliest plausible phase for limited, small-scale testing, not full deployment, with any meaningful constellation likely extending into the early 2030s as unresolved challenges—thermal engineering, radiation-hardened hardware, launch cadence, and orbital networking—are addressed.

Crucially, there is no published SpaceX roadmap with defined milestones, deployment phases, or capacity targets, at least we’ve seen publicly. Of course, Elon likely has detailed plans privately.

Chamath Palihapitiya

Source: Substack

The timeline rests on vision statements and preliminary regulatory signals rather than a disclosed engineering schedule. The takeaway for Rare Earth Exchanges™ readers is clear: this is a moonshot grounded in real technical logic, but Musk’s timeline should be interpreted as strategic signaling rather than a delivery commitment. Any downstream implications for power systems, permanent magnets, and critical mineral supply chains are therefore likely to emerge later, more incrementally, and far less smoothly than headline narratives imply.

What Checks Out: Hardware Is the Wall

Back to Mr. Palihapitiya’s macro diagnosis, a broadly persuasive one. AI scaling is increasingly constrained by physical limits—power generation, cooling, and capital intensity—well before algorithms exhaust their potential. Elon Musk has made the same point: software-centric thinking tends to underestimate hardware realities. Here, SpaceX is uniquely positioned. Its launch economics, manufacturing velocity, and Starlink’s widely cited profitability (often estimated around ~$8 billion annually, though not publicly audited) create an experimental platform few others can match. Folding xAI into that ecosystem plausibly extends runway for compute-heavy model development—an advantage most AI labs lack.

Where Assumptions Multiply

The orbital-compute vision hinges on several aggressive leaps:

Constraint	What Must Go Right (Technical Reality)	Probability of On-Schedule Execution (next five years)
Compute Density (~100 kW per ton)	Radiation-tolerant, high-performance silicon must achieve near-terrestrial performance, yields, and reliability while surviving launch stress, sustained radiation exposure, and long duty cycles in orbit.	Low–Moderate (25–40%) — feasible in labs and prototypes, unproven at scale within stated timelines
Thermal Management in Space	Heat must be dissipated via radiative surfaces with sufficient area and mass acceptable mass penalties; thermal design, not ambient cold, governs performance limits.	Moderate (40–55%) — solvable physics, but mass and complexity penalties likely slow deployment
Launch Cadence (Near-Hourly Starship)	Starship must achieve airline-like reliability, rapid reusability, and industrialized launch ops at a cadence with no historical analogue.	Low (20–30%) — aspirational within 3–5 years, unprecedented operational challenge
Latency & Orbital Networking	High-bandwidth, low-loss inter-satellite links must support distributed training and limited inference without cost or reliability blowouts.	Moderate (45–60%) — training workloads viable sooner than inference; networking remains a scaling risk

None of these barriers is impossible. All remain unproven at the proposed scale and may require more time to work out execution at scale.

The Quiet Constraint: Materials and Magnets

This is the missing layer Rare Earth Exchanges™ consistently flags. Orbital data centers would dramatically increase demand for permanent magnets, power electronics, and high-purity materials—supply chains still heavily dominated by China. Launch vehicles, satellites, solar arrays, and thermal systems intensify reliance on rare earths and other critical minerals.

Any claim of long-term cost leadership in space-based computing must therefore reconcile with magnet supply security, processing capacity, and price volatility. This unless Musk finds a way to decouple from rare earth technology—which he has expressed a goal of doing.  While the Trump administration has mobilized more federal attention—and loan and even equity-based capital—toward critical minerals than any prior U.S. administration, this effort still falls far short of a comprehensive, durable industrial policy.

Palihapitiya’s perspective carries a natural and understandable tilt: as a major rare-earth investor, he benefits from narratives that expand downstream demand. That does not invalidate his thesis—nor does it diminish the risk capital he has committed—but it underscores the need to interrogate upstream and midstream chokepoints with equal rigor. An impactful industrial policy is necessary if there is to be a full, sustainable, and scalable decoupling from China.  

Valuation Gravity

The implied $1.25 trillion valuation assumes simultaneous success across launch cadence, satellite manufacturing, AI performance, and regulatory clearance. Markets may reward the option value. Investors will distinguish visionary optionality from achievable technology lift in the near-to-intermediate term—and from bankable cash flows.

Bottom Line

The SpaceX–xAI thesis is directionally insightful, frankly exciting, and intellectually serious—squarely in Palihapitiya’s wheelhouse—and at the same time operationally speculative. Hardware matters. Space may help. Physics, materials, execution, and supply chains will ultimately decide this quest.

Source: Substack by Chamath Palihapitiya, Feb. 7, 2026.