Elon owns both sides now

In February 2026, SpaceX officially acquired xAI in an all-stock deal valued at $1.25 trillion, the largest private merger in history. SpaceX was pegged at $1 trillion, xAI at $250 billion. The transaction was structured as a share exchange, with xAI investors receiving 0.1433 shares of SpaceX for every share of xAI. On paper, it's a merger of a rocket company and an AI startup. In practice, it's the final piece in the most vertically integrated technology empire ever assembled. One person now controls the rockets, the satellites, the AI models, the social platform, the robotics company, and the electric vehicle fleet. When one person owns both the infrastructure and the intelligence that runs on it, the usual frameworks for thinking about competition stop working.

The full stack

To understand what this merger means, you have to look at the full picture of what Elon Musk now controls. SpaceX is the world's busiest satellite launch provider. It operates Starlink, a broadband mega-constellation with direct-to-device communications. xAI builds large language models, including Grok, and runs some of the world's largest AI supercomputers at its Colossus data centers. X (formerly Twitter) is a global social media platform generating vast amounts of real-time data. Tesla produces electric vehicles, collects enormous quantities of real-world driving data, and is developing humanoid robots with Optimus. No other entity on Earth has this combination. Google has AI and data but no rockets. Amazon has cloud infrastructure and is investing in Blue Origin, but its space capabilities are years behind SpaceX. Microsoft has AI partnerships through OpenAI but owns none of its own physical infrastructure stack. Apple controls hardware and software but has no presence in space or AI models. The combined valuation of $1.25 trillion puts this entity in the range of entire national GDPs. It would rank ahead of countries like the Netherlands, Switzerland, and Saudi Arabia.

Orbital data centers, the stated rationale

Musk's stated reason for the merger is to build orbital data centers. The idea is to launch AI compute infrastructure into space, where solar power is continuous and effectively free, cooling is handled by the vacuum of space, and there are no land constraints or community power grid impacts. SpaceX filed with the FCC for authority to launch a constellation of up to one million satellites dedicated to orbital compute. The company claims it can launch 100 gigawatts of AI compute capacity annually using Starship, with each flight eventually carrying 200 tons of specialized compute satellites. Musk has said he believes orbital AI compute will become cheaper than terrestrial alternatives within two to three years. The vision is ambitious, bordering on science fiction. Analysts at MoffettNathanson estimated the plan could cost trillions of dollars. Microsoft previously explored a similar concept with undersea data centers, Project Natick, and abandoned it due to cost concerns and the inflexibility of sealed, maintenance-free hardware. AI chips improve rapidly, and a satellite locked in orbit for five to seven years risks becoming obsolete long before replacement. But the point isn't whether orbital data centers will work on Musk's timeline. The point is that SpaceX is the only company that could even attempt it. It has proprietary launch capacity, operational satellite constellations, laser-based data transmission, and now in-house AI model development. If orbital compute does become viable, SpaceX has a monopoly on the pathway to get there.

The regulatory vacuum

The SpaceX-xAI merger closed remarkably quickly, largely because Musk controls both companies. There was no hostile negotiation, no competing bids, no extended regulatory review. Sullivan & Cromwell advised on the deal, and it was done. A legal analysis from Suffolk University's Journal of High Technology Law put it bluntly: the merger "does not obviously violate corporate law principles, antitrust rules, or regulatory requirements. Instead, it exposes the limits of those frameworks." The law allows this merger, but the law was not designed for a scenario where one individual owns the full AI supply chain from satellite to inference. Existing antitrust frameworks were built for horizontal monopolies, companies that dominate a single market. Standard Oil controlled 90% of U.S. oil refining when the Supreme Court ordered its breakup in 1911. AT&T controlled the entire telephone network when the Department of Justice forced its divestiture in 1984. Both cases involved a company that dominated one layer of one industry. The SpaceX-xAI entity doesn't fit these precedents neatly. It doesn't dominate any single consumer market in the way Standard Oil dominated refining. Instead, it controls multiple layers of multiple markets simultaneously: space launch, satellite internet, AI compute, AI models, social media data, autonomous vehicles, and robotics. The antitrust toolkit was designed to handle monopolies, not conglomerates that create their own competitive moats by connecting previously separate industries. Some commentators have argued that SpaceX should be subjected to common-carrier regulations, similar to how telecommunications infrastructure and railroads were historically regulated. The argument is straightforward: if SpaceX controls the only viable pathway to space-based compute, it should be required to offer access at publicly listed, non-discriminatory prices rather than giving preferential access to Musk's other businesses. So far, no such regulation has materialized.

The contradiction at the courthouse

The timing of the merger adds another layer. In the same period that Musk consolidated the largest private AI empire in history, his lawsuit against OpenAI and Sam Altman went to trial in Oakland, California. Musk's case alleges that OpenAI abandoned its nonprofit mission to benefit humanity, instead pivoting to a for-profit model that primarily enriched Altman and Microsoft. Musk claims the roughly $38 million he contributed to OpenAI was used for unauthorized commercial purposes. He's seeking an estimated $134 billion in damages and a change in OpenAI's management. During cross-examination, Musk described Altman and OpenAI co-founder Greg Brockman as "a deceitful pair" who steered OpenAI away from its altruistic roots. He testified that by late 2022, he had "lost trust in Altman" and believed they were "really trying to steal the charity." The contradiction is hard to ignore. Musk is suing OpenAI for abandoning its mission of building AI for the public good, while simultaneously building the most concentrated private AI empire on the planet. He argues that AI should not be controlled by a small group of people pursuing profit, while positioning himself as the single individual who controls more of the AI supply chain than anyone else in history. OpenAI's defense leans into this tension, arguing that Musk knew exactly what was happening and was primarily upset about not being made CEO. The jury will have to decide whether Musk's grievance is principled or personal. But regardless of the trial's outcome, the structural reality remains: one person now controls more of the AI stack than any nonprofit, government, or corporation.

The Apple precedent, and its limits

Vertical integration is not inherently bad. Apple proved that controlling hardware, software, and services together can produce genuinely better products. The tight integration between iPhone, iOS, and the App Store created an experience that no collection of separate vendors could match. But Apple's vertical integration has limits that Musk's does not. Apple doesn't control the cellular networks its phones run on. It doesn't own the chip fabrication facilities, it designs chips but outsources manufacturing to TSMC. It doesn't control the raw materials. And it faces genuine competition from Android, which powers the majority of smartphones worldwide. The SpaceX-xAI stack is different because at the infrastructure layer, there is no comparable competitor. No one else can launch satellites at SpaceX's scale or cost. No one else has a broadband mega-constellation. If orbital compute becomes the dominant paradigm for AI, SpaceX won't just be vertically integrated, it will be the vertically integrated gatekeeper of the entire industry. The question isn't whether vertical integration creates value. It clearly can. The question is what happens when the integrated entity also controls the infrastructure layer that everyone else depends on.

What this means for builders

If you're building an AI company, a SaaS product, or any technology business that depends on AI infrastructure, the SpaceX-xAI merger changes the calculus. Today, AI compute is primarily provided by cloud hyperscalers: AWS, Google Cloud, and Microsoft Azure. These companies compete with each other, which keeps pricing and access relatively fair. If orbital compute becomes viable and SpaceX controls the pathway, the competitive dynamics shift dramatically. Your AI provider's parent company could also be your competitor in every adjacent market. Tesla is integrating Grok AI into its vehicles and Optimus robots. SpaceX is integrating Starlink into Tesla vehicles. xAI uses Tesla's battery technology. The synergies are real and they create genuine product advantages. But they also mean that any company competing with Tesla in autonomous vehicles, or with xAI in AI models, or with Starlink in connectivity, faces a competitor that has preferential access to capabilities across the entire stack. This is the classic platform risk problem, amplified to a scale never seen before. When your infrastructure provider is also your competitor, the ground you're building on is not neutral.

Looking forward

The SpaceX-xAI merger is not a typical acquisition story. It's a structural shift in how technology power is concentrated. The combined entity doesn't just compete in markets, it defines the infrastructure those markets run on. Whether orbital data centers work on schedule is almost beside the point. The merger has already created an entity with capabilities no competitor can replicate. The regulatory frameworks that might constrain it were designed for a different era of technology. The legal challenges, including fiduciary duty breach claims from shareholders who saw their SpaceX stakes diluted, will take years to resolve. The precedents from Standard Oil and AT&T suggest that regulators eventually catch up to concentrated power. But "eventually" can take decades. Standard Oil operated for 29 years before the Supreme Court ordered its breakup. AT&T's consent decree came 75 years after the company was founded. In the meantime, we're watching something genuinely new take shape: a single individual with control over rockets, satellites, AI models, social data, electric vehicles, and robotics, building toward a future where AI compute runs on infrastructure only he can provide. It's not a company. It's a stack that spans from the ground to orbit, with one person at every layer.