Key Takeaways

  • SpaceX just landed a $45 billion, 3-year contract with Anthropic for "Elon Web Services" (EWS), an AI compute buildout that now rivals Starlink's revenue scale.
  • This EWS deal, valued at $1.25 billion a month, positions SpaceX as a critical infrastructure provider, capable of building data centers "dramatically faster... at a lower cost" than almost anyone else.
  • The new AI compute business is fueled by NVIDIA's urgent need for partners who can rapidly deploy GPU clusters, creating an unexpected and massive revenue stream for Musk's company.
  • Elon Musk's strategy centers on creating a "capital moat" by owning and deploying physical infrastructure, which then accelerates both technological and execution advantages in a self-reinforcing flywheel.
  • David Friedberg suggests space-based data centers offer a crucial "backup for civilization," ensuring information flow and human progress beyond the control of any single government.

"Elon Web Services" Is the New Starlink

Forget what you thought you knew about SpaceX's core business. While Starlink has been the talk of the town, a new, equally massive revenue stream has quietly emerged: "Elon Web Services" (EWS). This isn't just a side project; it's a $45 billion, three-year contract with Anthropic for AI compute, worth $15 billion annually. To put that in perspective, Jason Calacanis noted, "Anthropic is paying SpaceX... 1.25 billion a month to rent out Colossus 1 and parts of Colossus 2. It's a $45 billion deal over 3 years, 15 billion a year. In other words, they added a Starlink in terms of revenue to the party." This puts EWS squarely in the league of SpaceX's satellite internet venture, a shocking reveal for a business many didn't even know existed.

The implications are clear: SpaceX is not just launching rockets and satellites; it's becoming a foundational compute provider for the AI race. This makes the company a much bigger beast than most realize, driven by an unyielding demand for GPU capacity that only a few entities can satisfy at scale.

The Unseen Moat: Building What No One Else Can (or Will)

How does a space company become a top-tier AI compute provider? The answer lies in their unique ability to build. Gavin Baker explains it simply: “They build data centers dramatically faster than anyone else at a lower cost.” This isn't just about software optimization or clever algorithms; it's about the physical act of construction and deployment. In a world where NVIDIA is churning out GPUs, the real bottleneck isn't the chips themselves, but the ability to rapidly stand up the data centers to house and power them. SpaceX, with its intense engineering culture and rapid iteration, has unlocked this bottleneck.

Chamath Palihapitiya broke down Elon Musk's unique approach: “He creates a capital moat that then accelerates a technology moat that then accelerates an execution and a learning moat.” This isn't abstract strategy; it's about owning the physical means of production and deployment to a degree that creates an insurmountable advantage. For founders, this means looking beyond software layers to the core infrastructure that underpins your industry, and asking how owning or accelerating that could create a similar flywheel effect.

Space as the Ultimate Infrastructure Play

Beyond terrestrial data centers, the vision extends to space itself. David Friedberg raised a compelling point about the long-term strategic value: “I think the idea that you could have data centers, store information, transmit information, route information, and access information through space-based systems that can't be controlled, manipulated, or destroyed by governments, is is important.” This isn't just about convenience; it's about building a "backup for civilization," an unassailable infrastructure that ensures the free flow of information and human progress, even in times of extreme geopolitical tension. Chamath called it “probably the most important internet infrastructure project that's happened since the internet itself.” It positions SpaceX not just as a tech giant, but as a guardian of future human capability.

What to Do With This

This week, sit down and map out the biggest scaling bottlenecks in the AI market (or whatever high-growth market you're in). Ask: "What piece of physical infrastructure or execution speed is holding everyone else back?" Then, unlike most software-first founders, start brainstorming how you could own and deploy that critical physical layer faster and cheaper than anyone else. SpaceX found $45 billion in revenue by solving a problem NVIDIA and Anthropic couldn't solve alone: the physical act of building and deploying compute infrastructure.