It’s Christmas Eve in Sacramento.

Elon Musk’s jet makes an unannounced landing.

Onboard: Musk, his cousins, and a pair of wire cutters they recently purchased from Home Depot.

Musk had just bought Twitter.

He was on a mission to slash costs.

One target? A server farm engineers said couldn’t be shut down without weeks of prep.

Musk disagreed.

And when the engineers didn’t move fast enough…

Musk rented a truck, broke into the facility with his cousins, and started yanking out servers himself.

This is one of many stories I learned from William Isaacson’s biography on Musk.

Takeaway: Sometimes, the quickest way to “streamline operations” is to grab your own tools and get to work.

This, by the way, is exactly the mindset driving the rise of “edge computing” in 2025.

Yesterday, as you know, we talked about why we’re going to be hearing a lot more about “Edge AI” in 2025.

But you might be wondering:

If edge computing is set to revolutionize the AI landscape, reducing reliance on super AI farms…

What does that mean for mega AI centers like Colossus, Musk’s new baby in Memphis?

Glad you asked.

Inside Musk’s AI Empire

Colossus is a supercomputing powerhouse.

Packed with cutting-edge GPUs and increasingly unmatched processing capabilities…

It's the beating heart of Musk’s ambition to redefine AI.

On the surface, Colossus seems like the polar opposite of edge computing.

After all, Colossus is a centralized AI supercomputer—a mega-hive of GPUs designed to handle the heaviest, most complex workloads.

Edge computing, on the other hand, is all about pushing computation to the edges of the network…

Closer to the devices and users that need it.

But dig a little deeper, and you’ll see they’re two sides of the same coin.

Here’s why:

The Center vs. the Edge

Think of it this way:

Training happens in the center, execution happens on the edge.

Colossus isn’t designed to run your smart fridge or analyze data from a single drone in real time.

It’s designed to train massive AI models like the Grok family.

Think of it as a digital blacksmith forge—it creates the tools (AI models) that edge devices will eventually use.

Once trained on Colossus, these AI models can be deployed to the edge…

Where they’ll run locally on smaller, less powerful devices, reducing latency, improving performance, and minimizing bandwidth costs.

Example:

• Colossus trains an AI model that helps autonomous vehicles navigate complex environments.

• That model is then deployed to the edge, running on chips inside each car. The car doesn’t need to rely on a central server; it makes split-second decisions locally.

In other words…

Colossus Powers the “Brains”

Edge devices, by their nature, have limited resources—lower power, less compute, and smaller storage.

They rely on pre-trained models that need to be incredibly efficient and compact.

By leveraging Colossus'... well, colossal… processing power, xAI can develop and optimize edge-friendly AI models.

These are the brains behind smart cameras, IoT devices, and even robots in warehouses.

Without a supercomputer like Colossus, creating these models would take years—or be outright impossible.

(Decentralized alternatives are emerging to reshape AI training, but none rival the scale and power of Colossus.)

Where the Wild Money is Made

In short, Colossus and edge computing are part of the same ecosystem.

Colossus builds the AI that makes edge computing possible. The edge, in turn, feeds Colossus with real-world data to improve those AI systems.

It’s the ultimate feedback loop—centralized power driving decentralized intelligence.

The wild money is made in the middle: the infrastructure powering both.

If you’re an Altucher’s Investment Network member, you know our team has recently identified one company that stands to become one of the biggest winners in this next phase of the AI bull market.

More on that tomorrow.