A team of researchers from Google’s artificial intelligence labs has published results which it claims demonstrate that its controversial D-Wave quantum computer really works.

Google bought one of D-Wave’s quantum computers back in 2013. Claimed to be the “world’s first commercial quantum computer,” the device sits at NASA’s Ames Research Center in Mountain View, California, where it’s coded to tackle optimization problems that quantum computers are supposed to be good for. In theory, D-Wave’s hardware is supposed to be lightning fast—potentially 3,600 times faster than a supercomputer.

Quantum computers can, theoretically, be so much faster because they take advantage of a quirk in quantum mechanics. While classical computers use bits in 0 or 1, quantum computers use “qubits” that can exist in 0, 1 or a superposition of 2. In turn, that allows it to work through possible solutions more quickly.

But D-Wave’s chips have proved controversial. Teams of researchers have been unable to prove whether or not the computer actually takes advantage of true quantum effects.

Now, the team from Google has published results on the arXiv server that it claims demonstrate that the computer does indeed use quantum effects to solve problems at speed. In a series of experiments, the team pitted its D-Wave quantum computer against a regular computer with a single processor and had them race through optimization problems known as annealing.

The regular computer used a technique known as simulated annealing, while the quantum computer used what’s unsurpsingly known as quantum annealing. Google explains what happened:

We found that for problem instances involving nearly 1000 binary variables, quantum annealing significantly outperforms its classical counterpart, simulated annealing. It is more than 10^8 times faster than simulated annealing running on a single core. We also compared the quantum hardware to another algorithm called Quantum Monte Carlo. This is a method designed to emulate the behavior of quantum systems, but it runs on conventional processors. While the scaling with size between these two methods is comparable, they are again separated by a large factor sometimes as high as 10^8.

That looks like a convincing victory, with the D-Wave quantum computer completing the task 100 million times faster than the regular computer.

It may not quite be time to pop the champagne corks yet, though. First, the research is yet to be peer-reviewed. Second, as Technology Review points out, even if the results are verified, they overlook the fact that the regular computer wasn’t using the most efficient algorithm it could. It was using a technique that was algorithmically similar to the one running on the quantum computer, but an alternative approach could potentially have let it run through its calculations faster.

The Google team points out that the potential for the regular computer to run faster wouldn’t extend as the problems grew in size in the future. That’s why it chose to compare the two algorithms it did. But for now, the new claims still seem likely to be disputed by quantum physicists.

*Image by D-Wave*