SpaceX just became an AI company that also launches rockets.
The aerospace firm reported $14.6 billion in Q3 revenue on Tuesday, up 161 percent year-over-year. But the real story is in the breakdown. SpaceX’s “Super Intelligence” segment brought in $7.6 billion, up 830 percent from a year ago. That’s more than half the company’s total revenue, and it dwarfs the $2.4 billion operating income from Connectivity (Starlink) and the $7 million loss from the actual Space division.
SpaceX didn’t explain what “Super Intelligence” means or what it’s selling. But the numbers suggest Elon Musk has quietly built one of the largest AI businesses in the world while everyone was watching OpenAI and Anthropic. The company hit $4.4 billion in net income for the quarter, and AI was the engine.
The timing is notable. SpaceX announced that Tesla has formed special committees to discuss a possible merger. Musk said the committees were formed at shareholder request and consist of independent directors, and that he’s not involved in the discussions “because I’m a conflicted party.” The structure suggests serious conversations are happening, even if regulatory hurdles make a deal unlikely.
While SpaceX prints money from AI, the infrastructure layer is getting messier.
Google released a developer tool this week that makes it easy for PyTorch users to move their applications off Nvidia GPUs. Amazon did the same thing recently. Both companies are going after what’s become Nvidia’s most valuable asset: not the chips themselves, but the developer lock-in that comes with CUDA and the PyTorch ecosystem.
This is about margin and control. Cloud providers pay Nvidia massive amounts for GPUs, then rent them to customers at a markup. If they can shift developers to their own silicon, they capture more value and reduce dependency on a supplier that’s also competing with them through its own cloud services.
The tools matter because switching costs are real. Developers build on PyTorch, which runs best on Nvidia hardware. Google and Amazon are betting they can lower those switching costs enough to pull workloads onto their own chips (TPUs for Google, Trainium for Amazon) without developers having to rewrite their code.
Speaking of cloud providers: Anthropic is reportedly struggling to secure enough compute from partners who increasingly see it as a competitor. The AI lab has close partnerships with Amazon and Google (who’ve collectively invested billions), but those relationships are getting complicated as Anthropic prepares for a November IPO and the cloud giants build their own models.
The conflict is structural. Anthropic needs massive amounts of compute to train Claude. Amazon and Google are two of the only companies that can provide it at scale. But both companies also compete directly with Anthropic in the AI model business. Infrastructure spending is ballooning, and Anthropic needs to show investors it can secure supply without getting squeezed on price or access.
This is what happens when the AI stack vertically integrates. Everyone wants to own multiple layers. SpaceX builds data centers and trains models. Cloud providers invest in AI labs and build competing models. Nvidia makes chips and offers cloud services. The partnerships that made sense two years ago are now full of tension.
Otter.ai raised $320 million from General Catalyst’s growth fund for its meeting transcription tool. The company said it will use the money for enterprise growth, international expansion, and product development. Otter competes in a crowded space with Zoom, Microsoft Teams, and a dozen startups, but it’s carved out a niche with reliability and accuracy.
OpenAI acquired Orchestra Bio to strengthen its biotech business. The startup’s team and technology will work on Rosalind, OpenAI’s AI model for life sciences research, and the Rosalind Workbench tool that helps researchers plan experiments and analyze results. It’s part of OpenAI’s broader push into vertical applications where foundation models can create real economic value.
Meta’s Muse hit 10 million US downloads in 40 days, a milestone only 11 other apps have reached. New data shows Muse is retaining users at higher rates than rival AI chatbots, though the app faced a privacy issue this week when shared conversations exposed personal information. Meta suspended the sharing feature while it investigates.
Eric Schmidt reportedly owns about half as much of Anthropic as CEO Dario Amodei, according to documents related to the company’s upcoming IPO. Schmidt has been an active investor in AI companies since leaving Google, and his Anthropic stake is about to generate a significant return.
The pattern across these stories is convergence. SpaceX is an AI company. Cloud providers are chip companies. AI labs need infrastructure deals that make them vulnerable to their partners. The lines between sectors are gone, and the competition is just beginning.
One email at dawn. The five stories that mattered, with the bits removed and the meaning kept. Free, for now.