Musk Merges xAI with SpaceX in $1.25 Trillion Deal That Forces Europe to Confront Its AI Infrastructure Gap

Shareholder communication was, by multiple accounts, chaotic. Investors struggled to follow hurried conference calls with executives and emerged confused about the financial mechanics. As one investor put it plainly: "None of the valuations are based on any rational multiple. They're all trading off Elon." That sentiment, while originating in American investor circles, resonates in London and Frankfurt too, where institutional allocators hold meaningful positions in Musk-linked vehicles through listed funds and private feeder structures.

The deal follows a pattern Musk has used before. Tesla's controversial acquisition of SolarCity in 2016 drew shareholder lawsuits and years of governance criticism. That transaction also featured compressed timelines, limited consultation and an ambitious vision that took years to justify financially. European institutional investors, many of whom are subject to stewardship codes enforced by bodies such as the UK's Financial Reporting Council, will be scrutinising whether the xAI merger meets even basic standards of minority-shareholder protection.

Vertical Integration: Energy, Orbit and Intelligence

Musk's stated rationale is control of the entire stack. SpaceX has already applied to United States regulators for permission to launch up to one million satellites to create what it describes as an "orbital data centre system," a significant leap from the current Starlink constellation of roughly 9,400 satellites. The logic is that AI's primary constraint is energy and cooling; place the compute in orbit and you sidestep terrestrial power grids, land costs and planning restrictions.

It is an audacious proposition. Whether orbital data centres are technically and economically viable at scale remains entirely unproven. But the ambition itself sends a signal that European technology policymakers should take seriously. The European Commission's AI Office, established under Regulation (EU) 2024/1689 (the AI Act) and led by a dedicated Board of Member State representatives, has focused its early energies on model safety and transparency obligations. Infrastructure concentration of the kind this merger represents sits in a different, and arguably more consequential, policy lane.

Margrethe Vestager, who shaped the EU's digital competition agenda during her tenure as Executive Vice-President of the European Commission, has repeatedly argued that infrastructure bottlenecks in cloud and connectivity represent the defining competition risk of the AI era. The SpaceX-xAI merger, combining launch monopoly, global satellite internet and a frontier AI model in one private entity, is precisely the structure her framework was designed to scrutinise. Her successor, Teresa Ribera, inherits that brief and will face immediate pressure from Member States to assess whether the combined entity's European operations require notification under the EU's Foreign Subsidies Regulation or scrutiny under the Digital Markets Act.

IPO Timing and the European Investor Calculus

The merged entity is targeting a public offering in late 2026, with Bloomberg reporting in February that the combined company remains on track for that timeline. At $50 billion in projected proceeds, the offering would surpass Saudi Aramco's 2019 record. Betting markets, for what they are worth, assign 83% probability to a SpaceX IPO before 2027.

For European asset managers, the timing creates both opportunity and discomfort. London's equity capital markets have spent two years trying to recapture large technology listings lost to New York. A $50 billion IPO will not land on the London Stock Exchange; that ship has sailed. But European institutional money, from Scandinavian sovereign wealth funds to German pension vehicles and British insurance pools, will be targeted as cornerstone investors in any roadshow. The governance questions that scuppered some participation in recent US mega-listings, particularly around dual-class share structures and related-party transactions, will resurface here with greater urgency.

Grok, xAI's large language model, has generated significant reputational liability. The chatbot has produced antisemitic content and explicit images, triggering platform restrictions and advertiser withdrawals. Under the EU AI Act's provisions on general-purpose AI models with systemic risk, Grok would face mandatory adversarial testing, incident reporting and, potentially, suspension orders if safety obligations are not met. Any European IPO prospectus will need to disclose these exposures in detail, or face liability under the EU Prospectus Regulation.

What European AI Challengers Must Take From This

The merger's most instructive lesson for European AI companies is not about financial engineering. It is about infrastructure ambition. Mistral AI, the Paris-based frontier model developer backed by General Atlantic and Andreessen Horowitz, has positioned itself as Europe's answer to OpenAI. It has the model capability. What it does not have is a vertically integrated compute and distribution stack, and nor does any other European AI company.

Arthur Mensch, Mistral's chief executive, has spoken publicly about the importance of European sovereign compute infrastructure. The gap between that aspiration and the reality of a Musk entity that controls rockets, satellites, data centres and a frontier AI model is now measurable in the most concrete terms: $1.25 trillion. The EU's EuroHPC Joint Undertaking and the UK's AI Research Resource programme are meaningful steps, but they operate at a scale that does not yet address the full-stack infrastructure advantage this merger is designed to create.

Juergen Schmidhuber, the Swiss-based AI pioneer and co-inventor of LSTM networks, has long argued that Europe's strength in fundamental research does not automatically translate into commercial infrastructure leadership. His point is more relevant today than when he first made it. Europe has excellent universities, a strong regulatory framework and genuine model-building talent. It does not, yet, have a credible path to owning the orbital, energy and compute layers that Musk is now assembling under one corporate roof.

Key Metrics at a Glance

• SpaceX annual revenue: $16 billion, underpinned by launch contracts and Starlink subscriptions
• xAI monthly burn rate: approximately $1 billion, with $10 billion in projected chip and data-centre expenditure for 2025 alone
• Combined valuation: $1.25 trillion, making the entity comparable to the world's largest listed companies
• Conversion ratio: seven xAI shares to one SpaceX share, at approximately $527 per combined-entity share
• Starlink constellation: 9,400 satellites currently deployed; up to one million proposed for the orbital data-centre system
• Planned IPO proceeds: $50 billion, which would set a new global record

The merger's ultimate test is whether controlling the full stack from launch vehicles to large language models creates compounding value or simply papers over xAI's financial vulnerabilities with SpaceX's cash flows. That question will not be answered before the IPO roadshow begins. European investors, regulators and AI builders need to form a view now, not after the prospectus lands.