AI Is Going to Space — Is This the Next Data-Centre Frontier?

Google is about to take one of the strangest steps yet in the AI race.

Next week, Alphabet’s $Alphabet(GOOGL)$   Google plans to send its Tensor Processing Units into orbit as part of Project Suncatcher, an experiment exploring whether AI computing could eventually move into space. The prototype is scheduled to fly on SpaceX’s Transporter-18 rideshare mission. 

And no — Google isn’t putting a giant data centre in space just yet.

This first mission is basically a reality check.

Can AI chips survive radiation?

Can they handle extreme temperatures?

Can they be cooled in a vacuum?

Can satellites eventually communicate fast enough to work together?

Those are some of the questions Google wants to answer. 

The bigger idea is fascinating.

AI data centres on Earth are running into a very real constraint: power.

Google’s theory is that satellites in low Earth orbit could access near-continuous sunlight and potentially generate substantially more solar energy than comparable systems on Earth. Eventually, Google wants multiple satellites to work together using high-bandwidth laser links. 

If that sounds futuristic, it is.

Google itself describes this as a long-term research project, and experts cited by Reuters say commercial viability is still years away because of launch costs, engineering challenges and satellite manufacturing constraints. 

But here’s what caught my attention:

The AI infrastructure race is getting bigger than chips.

It started with GPUs and TPUs.

Then data centres.

Then electricity and grid capacity.

Now we’re talking about potentially putting the computing infrastructure itself beyond Earth.

And SpaceX is sitting right in the middle of this experiment as Google’s launch partner.

👀 Would you invest in the idea of space-based AI becoming a real industry?

A. 🚀 Yes — this could become a huge long-term infrastructure theme

B. 💻 Interesting, but terrestrial data centres still have the advantage

C. ⚡ Power/grid infrastructure is the better AI bottleneck to invest around

D. 🤯 Way too early — fascinating technology, questionable economics

The wildest part?

The next AI infrastructure boom might not happen entirely on Earth.