Your Next Netflix Binge Might Come From... Orbit?

We call it "The Cloud."

It sounds fluffy, weightless, and invisible. But here on Earth, the "Cloud" is actually a massive, windowless concrete building filled with thousands of hot servers, humming loudly and drinking up millions of gallons of water just to stay cool.

And with the rise of AI, these data centers are getting hungry. Like, really hungry.

I just watched a brand-new TED Talk by Philip Johnston, the CEO of Starcloud, and he dropped a statistic that actually made me gasp: To keep up with the projected growth of AI, the US alone would need to build 50 to 100 new nuclear power plants in the next few years.

That is... not going to happen.

So, where do we put all these computers if we run out of power and water on Earth? Johnston says the answer is obvious: We put them in space.

Why the "Cloud" Belongs in the Stars

In his talk, Let's Build AI Data Centers in Space, Johnston argues that space is actually the perfect environment for high-performance computing. Here is why moving our servers to orbit solves the biggest problems we face down here:

• Unlimited Power: On Earth, solar panels stop working at night or when it's cloudy. In orbit, if you pick the right spot, the sun never sets. You get 24/7 solar energy that is five times more intense than what we get through the atmosphere.

• The Ultimate AC: The biggest cost for a data center on Earth is cooling (servers get hot!). In deep space, the temperature is nearly absolute zero. You don't need water towers or fans; you just let the vacuum of space suck the heat away for free.

The "Starship" Factor

You might be thinking, "Tanisha, isn't it incredibly expensive to launch a supercomputer into space?"

It used to be. But remember my last post about Starship? That is the key unlocking this whole idea.

Because launch costs are dropping from $100,000/lb to ~$900/lb, we can now afford to launch massive, heavy structures. Johnston describes a future where we launch "Starship-sized" data centers equipped with solar arrays that stretch for kilometers.

These floating server farms would handle the heavy lifting for AI training, beaming the results back to Earth via lasers.

A Green Internet?

The coolest part of this vision isn't just the tech; it's the environmental impact. Right now, data centers account for about 2-3% of global electricity use, and that number is skyrocketing.

By moving the most energy-intensive computers off-planet, we stop fighting for resources on Earth. We stop draining local water supplies to cool servers. We stop burning coal to train AI models.

We essentially outsource the pollution to space, where the sun is free and the "air" is always cold.

It sounds like sci-fi, but Johnston is already building the first prototypes. The Cloud is finally going to where it belongs: the sky.

Check out the full talk to see the wild concept art for these orbital server farms: Watch: Let's Build AI Data Centers in Space | Philip Johnston (Note: Search for Philip Johnston on TED.com)