Google’s Quantum Breakthrough

In October 2019, Google shocked the world…

Announcing that it had achieved quantum supremacy – thanks to its quantum computing chip, Sycamore – a point at which a quantum computer can perform a task that is impossible for a classical supercomputer to solve in any reasonable amount of time.

Google’s Sycamore Chip | Source: Google

Google’s research back then claimed that it could solve a complex computational task in just 200 seconds – a task that would take the world’s fastest supercomputer at the time more than 10,000 years to solve.

It was a breakthrough of historical significance.

Or was it?

A Tech Kerfuffle

The announcement led to a kerfuffle between Google and IBM.

It was humorous at the time because IBM was running the world’s most powerful classical supercomputer – Summit.

IBM claimed that Google’s calculations were wrong. It outlined in its own research how Summit could complete the task in just 2.5 days, with “far greater fidelity” (i.e. far fewer errors, more accurate results). And because 2.5 days is a very reasonable amount of time, the team at IBM boldly proclaimed that quantum supremacy had not been met.

I can imagine the insults that were being uttered by both teams when referring to the others. The dust-up was very public and very notable… considering it was coming from two of the most well-known technology companies in history.

It didn’t help that the very measure that Google used to determine quantum supremacy was created by Google itself.

Google had created the random circuit sampling (RCS) benchmark as a way of evaluating quantum computing performance. Talk about circular logic: “We’ve achieved quantum supremacy because we said we achieved quantum supremacy by our own measure.” I’m not suggesting that the team at Google was wrong, it’s just that the technical bravado of the situation was funny.

RCS is a very complex computational task where a quantum computer randomly selects quantum gates to establish a quantum circuit, and then it runs that circuit multiple times and measures those outputs.

It’s a benchmark that doesn’t have much utility, other than for measuring quantum computing performance. And while the RCS benchmark is now used in the industry, some companies have their own measures of their quantum computing performance.

The disagreement was never settled, but it didn’t matter. Research and development in quantum computing continued at a reinvigorated pace following Google’s announcement.

But surprisingly, Google’s quantum computing division was quiet after the Sycamore announcement.

Most expected annual breakthroughs from what is now known as Google’s Quantum AI group…

But that didn’t happen.

And now, after four years of waiting, just yesterday, the big breakthrough was announced.

Willow

Willow…

Google Quantum AI’s next-generation quantum computing semiconductor.

Source: Google

As disappointing as the four-plus year wait has been, it was worth the wait.

Manufactured in Google’s own semiconductor manufacturing facility in Santa Barbara, California, Willow is a radical improvement over Sycamore.

And this time, there will be no debate over quantum supremacy – from anyone.

Using Google’s Random Circuit Sampling (RCS) benchmark, Willow can perform a computation in under five minutes that would take the world’s current fastest supercomputer – Frontier – 10 septillion years (10²⁵years) to accomplish. That’s:

10,000,000,000,000,000,000,000,000 years.

The above number exceeds the age of the universe. To be clear, that is not a “reasonable” time frame… hence, clearly, quantum supremacy.

Willow is designed with 105 quantum bits (qubits) as compared to Sycamore’s 53 qubits. The difference might not seem like much, but it is. With each additional qubit added to a quantum computing system, the computational power grows exponentially.

But it wasn’t just the added qubits that made Alphabet’s (Google’s) stock jump 10% in the last 24 hours. Note: that’s a remarkable move for a company worth $2.3 trillion.

One-Month Stock Chart of Alphabet (GOOGL)

It was the research that demonstrated that Willow’s quantum coherence – improved to 100 microseconds, compared to Sycamore’s 20 microseconds.

That’s a 5X improvement over the last four years.

Quantum coherence is the ability of a quantum computing system to maintain its quantum states over some time. The longer the coherence, the more complex problems that can be solved.

To put things in context, a microsecond is one-millionth of a second. So even 100 microseconds is a tiny sliver of time. But because quantum computers are so powerful, meaningful work can still be done within a split second.

And even stronger factor in Google’s share price spike was a breakthrough in quantum error correction (QEC), which has been an industry grand challenge. After all, if there are too many errors, the computations will not be accurate.

Source: Google

And this is where Google’s latest announcement gets really exciting.

In research published in Nature, Google demonstrates that as it increases the number of qubits in use on the Willow quantum chip, the error rate declines exponentially.

Google now operates below the critical quantum error correction threshold, paving the way toward a universal, fault-tolerant quantum computer.

But one of the most interesting developments about this breakthrough in error correction is a development that no one is talking about.

DeepMind Does It Again

In November, Google’s DeepMind division announced the development of AlphaQubit, its latest AI breakthrough.

AlphaQubit is designed to decode quantum computers, identify quantum computing errors, and correct those errors to improve the fidelity of a quantum computer.

DeepMind, again, brings a breakthrough to solving a grand challenge. It’s the secret behind Google Quantum AI’s latest announcement.

But before we get too excited, we have to take a step back at the big picture. From Google’s perspective, it has only reached its second milestone in its quantum computing roadmap.

Source: Google

While Willow is running below the quantum error correction threshold, it still has a long way to go to dramatically reduce error rates, create logical gates between qubits, and make a large-scale, fault-tolerant quantum computer.

To what end, we might ask?

Well, for those who love science fiction, a universal fault-tolerant quantum computer could discover the physics of a wormhole, which could lead to interstellar travel.

Or it could be used to develop a warp drive, capable of propelling a spacecraft at speeds faster than the speed of light.

And of course, some far nearer future applications could have profound impacts on human civilization.

Quantum computers will be used to deeply understand magnetism, which has a direct application for controlling and containing a nuclear fusion plasma in a fusion reactor.

Not to mention the applications for running complex machine-learning tasks on quantum computers, for the discovery and optimization of things like materials, molecular compounds, or new chemicals with commercial applications.

The future of quantum computing is a future whereby we are no longer limited by the lack of computational power to solve the world’s most complex problems. It’s a world where radical new scientific discoveries await.

We’ll just have to be a bit more patient until we get there.

Tomorrow, we’ll dig in a bit more on the quantum computing industry and what’s been going on the last few weeks.

Something has definitely been afoot, resulting in some interesting market action.

Stay tuned. More to come…

Jeff