Elon's Weekly X-Files: The Tweets That Matter

Starship Flight 14 launched today, carrying its first revenue payload into orbit: twenty-six of its own Starlink satellites. This is the moment Starship stops being an expensive experiment... and starts earning its keep. Last week, we talked about Starlink's ongoing milestones. This week, Starlink became Starship’s first paying customer, which is... efficient. The launch itself, which drew one point eight million live viewers, was just the start. Musk followed up by predicting Starship is only two to three years away from achieving hourly flights. A cathedral of stainless steel, launching every sixty minutes.

He also made a modest proposal for Starlink. He said it's "increasingly probable" that the satellite network will eventually carry a majority of Earth's internet traffic. At that point, he noted, Starlink would de facto become the internet itself. Meanwhile, on the software side, things are also accelerating. Grok, the xAI chatbot, is now integrated into Tesla vehicles. It allows for hands-free management of your inbox and calendar while driving. What could go wrong. Grok's user growth is apparently faster than anything SpaceX has ever seen, with over four hundred thousand weekly users and climbing. This growth is fueling some familiar optimism.

Musk claims his AI efforts, just three years old, could reach pole position in the AI race in about six months. He's also "cautiously optimistic" about having an AI model on par with a hypothetical GPT-6 in just two to three months. Cautiously. And finally, a comment that pulls it all together. After news that Google plans to launch its own AI hardware on a SpaceX rocket, Musk stated simply that the amount of compute in space will "obviously round up to one hundred percent." So what does it all add up to? A rocket launch, a software update, a few more impossibly ambitious timelines. It's all the same story.

The real theme this week is infrastructure. Physical, tangible, hard-to-build infrastructure. As Musk himself posted, "Hardware is hard." And his entire playbook is based on mastering it while everyone else is focused on software. You can see the pattern clearly this week. Starship is the heavy-duty delivery truck. Starlink is the global data network that the truck is deploying. And Grok is the intelligence designed to run on top of that network. This isn't just about building a better chatbot. It's about owning the physical rails the entire system runs on. When he says Starlink will become the internet, he doesn't just mean it will be the most popular provider.

He means it will BE the underlying physical layer for global data transmission. And when he points to compute moving into space—starting with Google's Tensor Processing Units on a Falcon 9 next week—he's showing you where the next generation of data centers will be built. In orbit. His main AI competitors, Anthropic and OpenAI, are six and ten years old, respectively. They had a massive head start on algorithms. But Musk is betting that his three-year-old AI project can leapfrog them by leveraging something they simply don't have: the world's most dominant rocket company. He isn't just building an application.

He's building the planet the application runs on. This week wasn't really about a single rocket or a new chatbot feature. It was a progress report on a project of vertical integration at planetary scale. For years, the race for artificial intelligence was framed as a software problem—a battle of algorithms. Musk is reframing it as a hardware and logistics problem. A battle for launch capacity, for global bandwidth, and for orbital real estate. So the question this week sets up isn't just who will have the smartest AI in six months. It's who will own the sky it lives in.