Today’s quantum computers are no more than experiments. Researchers can string together a handful of quantum bits — seemingly magical bits that store a “1″ and “0″ at the same time — and these ephemeral creations can run relatively simple algorithms. But new research from IBM indicates that far more complex quantum computers aren’t that far away.

On Tuesday, IBM revealed that physicists at its Watson Research Center in Yorktown Heights, New York have made significant advances in the creation of “superconducting qubits,” one of several research fields that could eventually lead to a quantum computer that’s exponentially more powerful than today’s classical computers.

According to Matthias Steffen — who oversees Big Blue’s experimental quantum computing group — he and his team have improved the performance of superconducting qubits by a factor of two to four. “What this means is that we can really start thinking about much larger systems,” he tells Wired, “putting several of these quantum bits together and performing much larger error correction.”

David DiVincenzo — a professor at the Jülich Research Center‘s Institute of Quantum Information in western Germany and a former colleague if Steffen — agrees that IBM’s new research is more than just a milestone. “These metrics have now — for the first time — attained the levels necessary to begin scaling up quantum computation to greater complexity,” he says. “I think that we will soon see whole quantum computing modules, rather than just two- or three-qubit experiments.”

Whereas the computer on your desk obeys the laws of classical physics — the physics of the everyday world — a quantum computer taps the mind bending properties of quantum mechanics. In a classic computer, a transistor stores a single “bit” of information. If the transistor is “on,” for instance, it holds a “1.” If it’s “off,” it holds a “0.” But with quantum computer, information is represented by a system that can an exist in two states

at the same time, thanks to the superposition principle of quantum mechanics. Such a qubit can store a “0″ and “1″ simultaneously.Information might be stored in the spin of electron, for instance. An “up” spin represents a “1.” A “down” spin represent a “0.” And at any given time, this spin can be both up and down. “The concept has almost no analog in the classical world,” Steffan says. “It would be almost like me saying I could be over here and over there where you are at the same time.”

If you then put two qubits together, they can hold four values at once: 00, 01, 10, and 11. And as you add more and more qubits, you can build a system that’s exponentially more powerful than a classic computer. You could, say, crack the world’s strongest encryption algorithms in a matter of seconds. As IBM points out, a 250-qubit quantum computer would contain more bits that there are particles in the universe.

But building a quantum computer isn’t easy. The idea was first proposed in the mid-80s, and we’re still at the experimental stage. The trouble is that quantum systems so easily “decohere,” dropping from two simultaneous states into just a single state. Your quantum bit can very quickly become an ordinary classical bit.

Researchers such as Matthias Steffen and David DiVincenzo aim to build systems that can solve this decoherence problem. At IBM, Steffen and his team base their research on a phenomenon known as superconductivity. In essence, if you cool certain substances to very low temperatures, they exhibit zero electrical resistance. Steffen describes this as something akin to a loop where current flows in two directions at the same time. A clockwise current represents a “1,” and counter clockwise represents a “0.”

IBM’s qubits are built atop a silicon substrate using aluminum and niobium superconductors. Essentially, two superconducting electrodes sit between an insulator — or Josephson junction — of aluminum oxide. The trick is keep this quantum system from decohering for as long as possible. If you can keep the qubits in a quantum state for long enough, Steffen says, you can build the error correction schemes you need to operate a reliable quantum computer.

The threshold is about 10 to 100 microseconds, and according to Steffen, his team has now reached this point with a “three-dimensional” qubit based on a method originally introduced by researchers at Yale University. Ten years ago, decoherence times were closer to a nanosecond. In other words, over the last ten years, researchers have improved the performance of superconducting qubits by a factor of more than 10,000.

IBM’s team has also built a “controlled NOT gate” with traditional two-dimensional qubits, meaning they can flip the state of one qubit depending on the state of the other. This too is essential to building a practical quantum computer, and Steffen says his team can successfully flip that state 95 percent of the time — thanks to a decoherence time of about 10 microseconds.

“So, not just is our single device performance remarkably good,” he explains, “our demonstration of a two-qubit device — an elementary logic gate — is also good enough to get at least close to the threshold needed for a practical quantum computer. We’re not quite there yet, but we’re getting there.”

The result is that the researchers are now ready to build a system that spans several qubits. “The next bottleneck is now how to make these devices betters. The bottleneck is how to put five or ten of these on a chip,” Steffen says. “The device performance is good enough to do that right now. The question is just: ‘How do you put it all together?’”