On vertical integration, orbital compute, and the quiet migration of systemic power from public institutions to private constellations.

The argument in brief: Three events in February 2026—the $1.25 trillion SpaceX‑xAI merger, the expansion of the sole‑sourced MILNET military satellite network, and Saudi Arabia’s $3 billion buy‑in via Humain—are not isolated headlines. They are load‑bearing pillars of a single structural shift: the migration of sovereign capability onto private, transnational, and increasingly orbital infrastructure. Through a series of individually rational procurement decisions, the state is becoming a tenant of the systems it once built. This essay traces how that inversion happened, what it means for citizens who now find themselves sub-tenants of a corporate substrate, and what concrete steps could begin to reverse the asymmetry before it becomes irreversible.

I. Three Events, One Architecture

On February 2, 2026, SpaceX announced the acquisition of xAI in an all-stock transaction valued at $1.25 trillion, creating the most valuable private company on Earth. Elon Musk, in a memo posted to SpaceX’s website, described the combined entity as “the most ambitious, vertically-integrated innovation engine on (and off) Earth, with AI, rockets, space-based internet” and the X social media platform.

Sixteen days later, Saudi Arabia’s Humain—a subsidiary of the kingdom’s trillion-dollar Public Investment Fund—disclosed a $3 billion investment in xAI’s Series E round, with its stake converting into SpaceX shares.

And threading through both announcements was a program most Americans have never heard of: MILNET, a 480-satellite military communications constellation sole-sourced to SpaceX and already supplanting the Pentagon’s prior, competitively awarded system.

These are not unrelated developments. They are the visible surface of a single structural transformation: the migration of critical state functions—military communications, intelligence analysis, data infrastructure—from public institutions onto private, commercially operated, and increasingly extraterrestrial platforms.

The thesis is simple to state and difficult to dismiss: through a cascade of procurement decisions, not through revolution, the state is becoming a subscriber to the infrastructure it once commanded.

II. The Vertical Stack: Rockets, Satellites, and the Logic Layer

To grasp the significance of the SpaceX-xAI merger, consider what it consolidates. Before February 2026, Musk’s companies were already entangled—Tesla and SpaceX each invested $2 billion in xAI, and xAI had absorbed X (formerly Twitter) in March 2025.

The merger creates a single entity that controls the rockets that reach orbit, the satellites that populate it, the AI models that process the data flowing through those satellites, and the social media platform that generates a continuous stream of real-time human behavioral data.

Musk framed the deal around orbital data centers, noting that global electricity demand for AI simply cannot be met with terrestrial solutions. Two days before the merger announcement, SpaceX filed with the FCC for authorization to launch up to one million satellites as a space-based computing constellation (FCC NGSO application, Docket No. 25‑340, DA‑25‑1018).

The Engineering Reality Check

Whether this million-satellite vision is achievable on any meaningful timeline deserves quantification. The FCC application included no satellite mass, no deployment schedule, and no cost estimate. SpaceX requested waivers of FCC rules §25.164 (milestone deployment obligations) and §25.165 (surety bond requirements)—the standard mechanisms that prevent companies from warehousing spectrum without building.

The application depends on SpaceX’s Starship rocket achieving routine reusable operation, a milestone not yet reached: as of February 2026, the vehicle’s test flights have included three consecutive explosions before partial successes.

Beyond propulsion, the technical barriers are substantial:

  • Thermal Management: Managing heat in the vacuum of space at million-node scale remains an open engineering problem.
  • High-Bandwidth Ground Relay: Connecting orbital compute with terrestrial users requires massive ground infrastructure.
  • Debris Mitigation: A constellation of this size would increase Earth’s active satellite population by roughly 6,800 percent.

Harvard astrophysicist Jonathan McDowell noted that as of January 2026, there were approximately 14,500 active payloads in orbit—9,555 of them belonging to Starlink. Tim Farrar, a satellite industry analyst, characterized the filing as “quite rushed” and likely a narrative device for the anticipated IPO.

However, the speculative nature of the million-satellite vision does not diminish its current structural significance. The filing functions as a regulatory land claim on orbital shells and Ka-band spectrum. In the history of infrastructure monopolies, the most decisive victories have often been won not by building first, but by filing first.

III. MILNET: The State as Subscriber

In June 2025, it was revealed that the U.S. Space Force had quietly contracted with SpaceX for a dedicated military satellite communications constellation designated MILNET. The program comprises over 480 satellites designed to provide high-bandwidth, resilient communications for global military operations.

The Procurement Bypass

The contract is funded by the Space Force but administered through the National Reconnaissance Office, leveraging an existing classified Starshield contract. This allowed the Pentagon to bypass the multi-year competitive solicitation that typically accompanies programs of this scale.

Senator Chris Coons of Delaware was pointed at a June 2025 hearing: “That work, which has been going on for several years and had robust competition and open standards, has been replaced by something called MILNET, which is being sole sourced to SpaceX”. The Space Force requested $277 million in FY2026 for MILNET research and development.

The Proprietary Mesh

MILNET uses SpaceX’s Starshield terminals, configured for military-grade encryption and NSA HAIPE compliance. The design makes the network highly resistant to electronic warfare—but deeply integrated with SpaceX’s commercial infrastructure.

The model is “government-owned, contractor-operated,” but in practice, SpaceX designs, manufactures, launches, and operates the constellation. The government’s role is oversight, exercised through a mission director embedded within a commercial workforce. This is not the traditional defense-industrial relationship in which the state specifies requirements; it is one in which the state adapts its operations to a commercial platform.

IV. The Pentagon’s Cognitive Layer

In December 2025, the DoD announced that xAI’s Grok models would be deployed on GenAI.mil at Impact Level 5—the security tier for handling controlled unclassified information.

In January 2026, Defense Secretary Pete Hegseth announced Grok’s enablement on classified Pentagon networks, with data from military intelligence databases fed into commercial AI systems for analysis. The Pentagon’s third AI-acceleration strategy omitted any reference to ethical AI use and banned deployment of models incorporating “ideological tuning”.

The result is a layered dependency. MILNET is the transport layer (the channel). Grok on GenAI.mil is the application layer (the cognition). Both layers now share a corporate parent. When the channel and the cognition belong to the same entity, the state’s operational independence narrows to whatever margins that entity’s architecture permits.

V. The Transnational Shareholder: Saudi Arabia’s Orbital Bet

On February 18, 2026, Humain announced its $3 billion investment in xAI’s Series E round. Humain’s shares converted into SpaceX equity—approximately 0.24 percent of the combined entity.

The investment builds on a November 2025 partnership to jointly develop over 500 megawatts of AI data center infrastructure in Saudi Arabia. This is a resource swap: Saudi Arabia possesses surplus solar energy and capital; SpaceX-xAI possesses the orbital infrastructure and models that require them.

The geopolitical implications compound because sovereign investors of this scale routinely receive informational privileges—briefings and strategic roadmaps—that would be extraordinary in a defense context. SpaceX and the government did not respond to press inquiries about whether the Humain investment triggered CFIUS review, revealing potential gaps in the existing framework for scrutinizing foreign sovereign investment in defense-critical operators.

VI. The Jurisdictional Gap—and What Could Fill It

The Outer Space Treaty of 1967 assigns jurisdiction to the launching state and requires “authorization and continuing supervision” of private space activities. In practice, this architecture is buckling.

Legal scholars have noted that enforcement mechanisms put the responsibility on states rather than private companies, leaving private actors mostly unregulated. The treaty’s key regulatory verb—“continuing supervision”—has never been operationalized for commercial mega-constellations.

This is a governance gap, not a void. Tools exist; they require political will and institutional imagination to sharpen.

VII. What Must Be Done

1. Congressional mandate for open-standards interoperability

Congress should require that any military communications constellation above 100 satellites maintain interoperability with at least one alternative provider’s terminals. Sole-source justifications for programs of this magnitude should be subject to mandatory unclassified reporting. Conditioning the $277 million annual MILNET appropriation on open-architecture compliance is the simplest first step.

2. Establishment of an Orbital Infrastructure Review (CFIUS expansion)

An explicit statutory mandate should trigger review based on operational sensitivity, not ownership percentage. Any company operating classified satellite constellations or national security launch contracts should be subject to mandatory foreign investment disclosure and review.

3. A multilateral initiative for space-based computing norms

The absence of international norms invites “regulatory arbitrage”. We need a transparency regime for defense-integrated constellations, using the Wassenaar Arrangement as a partial model to define threshold capabilities for orbital compute.

VIII. The View from Below: What a Tenant State Means for Its Citizens

A military contingency: A NATO ally requires urgent intelligence relay; the data transits MILNET; the targeting analysis runs through Grok. The state commands; the substrate determines whether to comply, and at what speed.

A civilian emergency: After a hurricane, emergency responders and hospitals rely on the same commercial mesh that interconnects with MILNET. The sovereign’s emergency toolkit is leased, not owned.

If the state is a tenant of private infrastructure, the citizen is a sub-tenant—further removed from the systems of accountability. The “revocable charter” of the 18th century (e.g., the East India Company) has been replaced by proprietary firmware in orbit. If the state cannot audit the code on a satellite it does not physically control, the power asymmetry is architectural and potentially irreversible.

IX. What This Isn’t—and What It Is

It would be analytically lazy to cast this as a story of a single villain. This consolidation is the result of genuine technical achievement, chronic underinvestment in public-sector capacity, and a global AI competition that rewards speed over deliberation.

But accepting the trajectory as inevitable is democratically negligent. The pattern to watch is the continued accretion: each new contract and each new sovereign investor. Collectively, they constitute a parallel infrastructure—private, transnational, and increasingly indispensable.

The architecture is being assembled overhead. Those who wish to preserve the possibility of self-governance would do well to look up—and then to act, while the tools of democratic accountability still have something to grip.

Sources and Publication Notes

The facts cited in this essay are drawn from public reporting by CNBC, Bloomberg, TechCrunch, CNN, Breaking Defense, SpaceNews, Via Satellite, Defense One, DataCenterDynamics, The Register, Broadband Breakfast, Reuters, Inside Defense, Analytics Insight, the Motley Fool, and official statements from SpaceX, xAI, Humain, the FCC (DA-25-1018), the U.S. Space Force, and the DoD.

  • FCC Filing: Space Exploration Holdings, LLC, DA‑25‑1018, Docket No. 25‑340 (accepted Feb. 4, 2026).
  • MILNET Data: Breaking Defense (June 18, 2025); Via Satellite (June 26, 2025).
  • Humain Investment: CNBC (Feb. 19, 2026); Humain press release (Feb. 18, 2026).
  • GenAI.mil & Grok: DoD press release (Dec. 22, 2025); Defense One (Jan. 13, 2026).