¿Las ambiciones espaciales de Altman: una distracción peligrosa?

The battle for the soul of Artificial Intelligence has officially left the planet.

For most of 2025, the rivalry between Sam Altman (OpenAI) and Elon Musk (xAI) was fought with silicon chips and hiring offers. But startling new reports confirm that Altman attempted to open a new front in the war: Low Earth Orbit.

Altman reportedly spent months exploring an acquisition of Stoke Space, a reusable rocket startup founded by former Blue Origin engineers. While the deal fell through, it reveals a critical pivot in the AI arms race. This isn’t just about billionaire egos; it’s about the physical limits of energy on Earth and the sudden emergence of Google’s Project Suncatcher.

The Stoke Space Play: More Than Just Rockets

To understand why an AI CEO wants a rocket company, you have to look at what Stoke Space is building. Unlike legacy aerospace firms, Stoke is designing a fully reusable second stage—the “Holy Grail” of rocketry that even SpaceX is still perfecting with Starship.

Why Altman Wanted It:

1. Energy Independence: AI models are consuming gigawatts of power. Earth’s grids are struggling. Space offers 24/7 unfiltered solar energy.
2. The “Musk Hedge”: Currently, if you want to put something big in orbit cheaply, you have to pay Elon Musk. For Altman, being dependent on his arch-rival’s infrastructure is a strategic non-starter.
3. The Talent: Stoke’s team is comprised of top-tier engineers who defected from Blue Origin and SpaceX.

Acquiring Stoke would have given Altman his own “FedEx for Orbit,” allowing OpenAI to deploy custom infrastructure without tithing to the Musk empire.

Google’s “Project Suncatcher”: The Elephant in Orbit

While Altman was shopping for rockets, Google was reportedly finalizing the designs for Project Suncatcher.

Unveiled quietly in November 2025, Suncatcher is a “moonshot” initiative to deploy autonomous data centers equipped with specialized TPUs (Tensor Processing Units) into stable orbit. The logic is terrifyingly simple:

• Cooling: Space is cold. You don’t need water or AC to cool your chips.
• Power: Solar panels in orbit generate power 24/7 with zero night-time intermittency.
• Latency: For non-real-time training runs (which take months), latency doesn’t matter. You upload the job, wait a month, and download the weights.

Google’s ability to execute this internally—leveraging its own chip design and massive capital—threatens to leave OpenAI earthbound. Altman’s move for Stoke was likely a defensive reaction to this existential threat.

Analysis: A Dangerous Distraction?

The question remains: Can the CEO of OpenAI afford this split focus?

Running a frontier AI lab is a full-time war. Google’s Gemini team is shipping updates weekly. xAI is burning cash to build the “Colossus” supercluster.

• The Bear Case: Building rockets is notoriously the hardest business in physics. If Altman had bought Stoke, he would be splitting his brain cycles between solving AGI alignment and solving turbopump thermodynamics. That is a recipe for losing both wars.
• The Bull Case: Compute is the bottleneck. If energy constraints on Earth cap AI growth, then the company that owns the “Space Cloud” wins the century. Altman isn’t distracted; he’s looking three moves ahead.

The Verdict

The deal for Stoke Space may be dead, but the intent is clear. The era of “AI companies” is over. We are entering the era of “Civilization Builders,” where the leaders of tomorrow need to own the full stack—from the code that thinks, to the chips that run it, to the rockets that launch it.

Altman dodged a bullet by not buying a rocket company today. But don’t be surprised if OpenAI creates a “Space Operations” division tomorrow.