SpaceX and Tesla are discussing a merger, and Elon Musk has framed cooperation on superintelligence software as one reason the companies could benefit from joining forces. Their businesses already touch artificial intelligence, data centers, self-driving technology, robots and chip development. The proposal also raises practical questions about finances, unfinished projects and how much of SpaceX’s reported AI revenue comes from its own products.
A shared technology vision
Musk’s stated ambitions for the companies connect work on Earth with plans for space. He has described going to Mars and spreading “the light of consciousness” as goals. The logic presented for a merger is that autonomous machines could help with exploration: communication delays between Earth and the Moon or Mars would make it difficult to control robots continuously from here.
In that scenario, robots would need to make decisions with limited supervision. The source article connects that possibility to Tesla’s self-driving cars and humanoid robots, and to language models and data centers associated with SpaceX. Musk said on a recent SpaceX earnings call that “there’s a lot to be gained from cooperating on developing superintelligence software.”
That is a forward-looking rationale, though. Whether readers find it persuasive depends in part on how they view Musk’s stated space ambitions and the practical role of AI in them. A merger could also bring financial and control benefits to Musk, independent of whether the larger Mars vision comes to pass.
What SpaceX’s AI revenue may represent
SpaceX reported $7.56 billion in revenue from its artificial intelligence business in its Q3 earnings. Within that, “SI solutions & infrastructure revenues” reached $7 billion, more than three times the prior quarter. The category reportedly includes Grok, Cursor and the neocloud business, although the article notes that the breakdown was not fully disclosed.
Cursor contributed $608 million in revenue after being added to the business. Analyst Gil Luria of DA Davidson said most of the remaining $7 billion was likely from renting computing capacity to Anthropic and Google, with Grok contributing less. The company’s reported figures therefore do not establish how much revenue comes from its own AI software versus infrastructure services.
The article contrasts this mix with Anthropic’s frontier-model revenue. SpaceX’s AI business may look large at first glance, but the comparison can obscure a difference in what the companies sell: SpaceX’s reported category includes infrastructure revenue, while Anthropic’s cited figure is tied to its frontier models. The release did mention increased Grok adoption among enterprise customers and consumers, but did not break that revenue out separately.
Tesla’s priorities and joint projects
Tesla’s Q3 results followed on Wednesday, with a second consecutive quarter of declining profits. The article describes Musk as less focused on conventional cars and more focused on robots and self-driving taxis. The companies already have some financial ties: SpaceX bought $131 million of Cybertrucks from Tesla and had previously held Tesla bonds.
They have also announced two joint AI-related efforts. Terafab is a planned chip-making venture in Texas intended to supply chips for Musk’s companies, with Intel involved in developing the technology. Macrohard is described as an AI project intended to simulate software already made by Microsoft. The article notes that little had been heard about Macrohard since its announcement in March.
The plans remain unsettled. SpaceX’s S-1 from May said the companies had not finalized details such as financial terms, intellectual property rights or the collaboration’s duration. It also said there might never be a definitive agreement. Intel’s CEO said the company remained committed to the partnership, but Intel was not obligated to stay involved.
Timelines and unanswered questions
On the SpaceX investor call, Musk said Terafab would reach volume production in the second half of 2028, with a first test chip expected in May or June 2027. When an analyst asked whether Musk had ordered the equipment needed to make advanced chips, he did not answer. Those dates are plans, not confirmation that the facility or production schedule is secured.
The source article also points to a gap in public reporting: neither Terafab nor Macrohard appeared in the companies’ latest 10Q, while the S-1 described the projects as being at a very early stage. That does not prove no work is happening, but it leaves investors without finalized public details on important parts of the proposed cooperation.
A merger could unite companies already linked by purchases, shared projects and overlapping technology ambitions. The case for it depends on whether those pieces can turn into working products and durable business value. For shareholders, the unresolved terms, the source of SpaceX’s AI revenue and the progress of Terafab are central to judging the proposal.