Tech moves fast, but you're still playing catch-up?

That's exactly why 200K+ engineers working at Google, Meta, and Apple read The Code twice a week.

Here's what you get:

All delivered twice a week in just 2 short emails.

Join 200K+ engineers

Have you ever “donated” your computer to science research? There was a time, not too long ago, when people would voluntarily donate their computer’s idle processing power to scientific projects. The famous ones are helping analyze radio signals for extraterrestrial life or simulate protein folding (links below).

It was a simple but powerful idea: one machine might be limited, but millions of connected machines could become something far greater.

That same philosophy quietly evolved into what we now call cloud computing. Instead of relying on a single powerful computer, companies began building facilities packed with thousands (or millions) of servers. These are data centers: the beating heart of the modern digital world.

At its core, a data center is a highly organized collection of computing resources. Rows of servers (essentially powerful computers) are stacked in racks, connected through high-speed networks, and supported by massive cooling and power systems. These facilities are designed for reliability, scale, and efficiency.

When you stream a movie, train an AI model, or store files in the cloud, you’re not using your local machine, you are tapping into these remote infrastructures. Modern AI, in particular, is inseparable from data centers. Training large models requires enormous computational power, often distributed across thousands of GPUs working in parallel.

The trick behind this scalability is distribution. A large problem is broken into smaller pieces, each handled by different processors. The results are then combined to produce the final outcome. This is how supercomputers and cloud platforms achieve performance that no single chip could ever deliver.

This distributed model has been wildly successful. Supercomputers are essentially vast networks of interconnected processors. Data centers extend that idea globally, allowing anyone with an internet connection to access immense computational power.

So naturally, the question arises:

Can we scale quantum computing the same way?

In the last episodes (check here), we’ve seen a variety of approaches to building quantum computers—using atoms, trapped ions, superconducting circuits, and even photons. Each architecture offers unique advantages, but they all face a common challenge: scalability.

Adding more qubits is not like adding more transistors on today’s semiconductor chips. Qubits are fragile. They lose information quite easily, require extreme environmental control, and become exponentially harder to manage as their number grows.

At first glance, the solution seems obvious: do what classical computing did. Connect many smaller quantum processors, often called Quantum Processing Units (QPUs), into a larger system.

And indeed, this is the idea behind distributed quantum computing and quantum data centers.

But here’s the catch.

In classical systems, dividing a problem is straightforward. You can split a dataset, assign chunks to different processors, and recombine the results later. The processors don’t need to share anything more than data.

Quantum computing doesn’t always allow that.

Some quantum algorithms rely on entanglement, a uniquely quantum property where qubits become deeply interconnected. In such cases, you can’t simply separate parts of the computation. You may need to perform operations, quantum gates, between qubits that reside on entirely different QPUs.

This is where things become fundamentally challenging.

To enable this, quantum systems must:

In simpler terms, you cannot always break a quantum problem into completely separate parts. Sometimes, different quantum processors need to “talk” to each other to finish the job. So you do need to connect the quantum processors, but those connections are not perfect. They’re noisy and fragile, which means every time processors communicate, there’s a risk of errors creeping in. As you add more connections, this noise can build up, ultimately reducing the overall efficiency of the computation.

So the goal isn’t to connect everything to everything else. That would be too complex and inefficient. Instead, we connect processors in a smart and limited way, just enough so they can communicate when needed. This is called a so quantum network!

A good way to think about it is a railway system. You don’t build a direct train line between every pair of cities, that would be too expensive and unnecessary. Instead, you create key routes and hubs that let people travel almost anywhere with a few connections.

Quantum networks work in a similar way: carefully chosen connections so the system stays efficient while still being powerful.

As you might have realized, moving from a single quantum processor to a network of them introduces multiple layers of optimization problems. One of the key challenges is how to split a quantum circuit in the most efficient way so it can run on a distributed quantum computing architecture. There is active research in this direction, for example, see the recent work below:

Despite the hardware challenges, the idea of quantum data centers is gaining traction. Imagine a facility similar to today’s data centers, but instead of just classical servers, it houses multiple quantum processors connected through quantum networks.

These centers would combine:

This distributed approach addresses two critical challenges:

In essence, quantum data centers could do for quantum computing what cloud platforms did for AI.

But the challenges are very real and go beyond straightforward engineering scaling. Entanglement distribution is extremely sensitive and error-prone, communication losses can permanently destroy quantum information, and even small imperfections in timing can disrupt synchronization across nodes that must operate in near-perfect coordination. On top of this, error correction becomes even more complicated in a distributed setting, since it must now account for both local hardware noise and the additional noise introduced by inter-processor communication.

The story of computing has always been about scaling, finding ways to go beyond the limits of a single machine. Data centers and supercomputers solved this for classical computing through distribution.

Quantum computing is now facing its own version of that challenge, but with entirely new rules. Many companies are now racing to build what could become the world’s first quantum data center!

It’s a harder problem. But if solved, it could unlock a level of computational power that no standalone machine could ever achieve.