The Quantum Takeover Just Got A Lot Closer

“The end of the decade.” What does that sentence mean to you? In your head you would probably think to go to the next decade after 2020, and eventually come to 2030. Now, what if I told you four years? In theory, that would mean a lot more to you because now instead of talking decades we are talking years.

In theory, quantum computing will be everywhere in four years, in 2030. No, I’m not talking about another quantum chip, I’m talking about operational quantum computers all over the world. Well, you might be wondering, “don’t we already have quantum computers in very distinct areas being worked on?” And the answer to that question is, yes, yes, we do. But, there’s a crucial distinction: there’s a difference between a car that can start its engine and a car that can drive on the highway.

So, what exists today?

Quantum computers today are NISQ machines (Noisy Intermediate-Scale Quantum computers). In fact, quantum computers are so error-prone that some classical laptops can beat them. However, the goal of the current quantum computers is the knowledge of the potential of where they can be. Engineers today know that quantum computers have the potential to run miles around today’s fast classical computers, and because of that known potential they continue to develop them.

What will exist in 2030?

What Caltech is claiming: in 2030, the first fault-tolerant quantum computer will be created. What I mean by fault-tolerant is that it will be able to correct its own errors, and outperform all classical computers on proper tasks.

Let’s turn back seven months, going back to August 1, 2025. Fujitsu Limited, based in Kawasaki, Japan, announces that it has started research and development towards a superconducting quantum computer with a capacity of 10,000+ qubits.

Fujitsu’s announcement was nothing short of mind-blowing, thinking about a 10,000 qubit quantum computer being here in 2030 just makes you start wondering: are we even going to have classical computers then? Anyways, Fujitsu’s choice of 10,000 qubits was purely ambition, and the fact that bigger machines means more complex computations. The project was well-funded, but it was still a bet. Nobody proved that 10,000 was the number of qubits needed to be used universally. The horizon wasn’t defined, and the goal was just a surplus.

Then, seven months later, Caltech defined the horizon for them.

In March 2026, a team at Caltech showed that a fault-tolerant quantum computer capable of running Shor’s algorithm could be built with as few as 10,000 qubits. But being capable of running Shor’s algorithm isn’t just being able to run a complex system, it’s having the capability to destroy modern encryption. Shor’s algorithm has been laid out for thirty years now, and the reason it wasn’t being implemented is due to us not having the resources or the platform to run it. Caltech just changed that, it’s here in four years.

One week before the Caltech paper dropped, Google Quantum AI announced that it was expanding its quantum computing system to include neutral atom computing alongside its superconducting research. Superconducting systems are efficient, but neutral atom arrays can scale much better for a larger number of qubits. This means that Google is chasing this goal as well, and it was acknowledging that neutral atoms solve problems that superconducting can’t, and that fault-tolerant quantum needs both.

But one issue with superconducting quantum (which is what Fujitsu is using) is that no two qubits are perfectly identical. Additionally, they require cooling to -273ºC, making it very expensive to regulate. The reason that this issue needs solving is because currently superconducting quantum is noisy, and the qubits have very short coherence time.

This is where research in labs starts assisting these companies. Professor Yao Wang’s lab at Emory has been applying machine learning directly to quantum materials, which are training models to read complex measurement signals and identify patterns. This approach finds that instead of treating every qubit the same way, ML can learn the specific issues of each individual qubit and tailor the algorithm towards that. Error-correction currently tries a one-size-fits-all solution, but this becomes personalized. The imperfections aren’t mitigated, what’s changed is the impact on the algorithm.

In the last two weeks, quantum has reminded us that the shift is coming, and the world must adapt…quick. I mean your bank account, messages, medical records, and practically every piece of personal digitized information is at risk from Q-Day (the moment a quantum computer breaks modern encryption), and we must be prepared for the switch by 2030. The hardware roadmaps are real, the predictions are real, quantum isn’t just a distant theory anymore.

“The end of the decade” just got a lot closer than it sounds.