Here’s what you’ll learn when you read this story:

• Current quantum computers don’t possess enough qubits to crack classical encryption methods, but a flurry of new research suggests that the threshold for such a breach could require drastically fewer qubits than previously estimated.
• Until recently, researchers thought that 20 million qubits would be needed to crack schemes like elliptic-curve cryptography (ECC), but a new preprint from Caltech revises that number down to just 10,000.
• This improved performance is mostly derived from improved error correction through non-local communication, which increases a quantum computer’s fault tolerance.

The forthcoming era of quantum computers holds a lot of promise. Qubits—the quantum version of classical bits—have the potential to solve immensely complicated problems that today’s computers could never hope to tackle, thanks to their ability to leverage the quantum mechanical properties of superposition and entanglement. But with great power comes great responsibility, and these quantum technologies also have the dangerous ability to break classic cryptographic schemes like elliptic-curve cryptography (ECC, which is the backbone of cryptocurrencies like bitcoin) and 2048-bit RSA (one of the oldest public-key cryptosystems), which ensure the security of our online lives.

Until relatively recently, researchers estimated that quantum computers likely needed at least 20 million qubits in order to break through these types of cryptosystems, but a new study by researchers at Caltech has drastically revised those numbers down to as low as 10,000 qubits. At the moment, no quantum computer is close to this number—Caltech recently revealed its 6,100-qubit array in late 2025, and most commercial quantum computers hover around 1,000 qubits max. But the newly revised estimate—detailed in a paper uploaded to the preprint server arXiv— means that an encryption apocalypse is probably closer than we thought.

At the heart of this breakthrough is a new quantum error-correction architecture that’s way more efficient than what was possible in previous machines. The problem with qubits is they can be faulty (largely due to thermal noise or decoherence), but extra qubits and careful construction can correct for these quantum failures and eventually make machines “fault-tolerant.” In the 6,100-atomic qubit array, for example, Caltech scientists used laser beams called “optical tweezers” to precisely arrange qubits, enhancing entanglement. According to Ars Technica, the benefit of this arrangement is that each qubit participates in “non-local communications,” meaning that all physical qubits can interact with all other qubits.

“Unlike other quantum computing platforms, neutral atom qubits can be directly connected over large distances,” Manuel Endres, a co-author of the paper from Caltech, said in a press statement. “Optical tweezers can shuttle one atom to the other end of the array and directly entangle it with another atom.”

The researchers estimated that a quantum computer with this new architecture would require only 9,998 qubits to crack ECC in 1,000 days, and just 26,000 qubits to do so in a single day. As for RSA-2048, the machine would need around 100,000 qubits and approximately 10 days. Another paper published by Google earlier this week estimated that 500,000 qubits could render modern encryption techniques obsolete in just a few minutes. All of these methods require quantum computers with numbers of qubits far beyond what’s technically feasible today, but the estimates still represent drastic reductions from the many millions researchers formerly believed would be necessary.

“I always considered theoretical research on the usefulness of large-scale quantum algorithms to only be of interest in the distant future,” Hsin-Yuan, a co-author of the study from Caltech and CTO of Oratomic (a quantum computing start-up), said in a press statement. “Our new study made me realize they might come true in the next few years.”

But even if these quantum computers eventually do arrive, the world’s digital security infrastructure won’t suddenly be at the mercy of a shadowy group of quantum hackers. Researchers are simultaneously investigating ways to develop quantum cryptography that can confound even the most advanced of quantum computers.

For decades, the quantum cryptographic arms race was a theoretical one, but in the next few years, it could become very, very real.