with CLIPREVIEWED learn the articleA Bubble Ready to Burst?
The massive intersection of James Avenue and East Boughton Road in Bolingbrook, Illinois, looks like many other crossroads in suburban America. A drive-through Starbucks keeps watch over 15 lanes of turning and merging mid-size SUVs, most headed for the sprawling parking lots of the Promenade shopping mall to the south, a few others en route to the shooting gallery and gun shop across Interstate 355 to the east.
Few of the people in the SUVs realize they’re driving over part of America’s blossoming research into quantum information technology. Beneath the interstate, entangled photons—quantum particles moving at the speed of light—are teleporting to and from the Argonne National Laboratory in the next town over, through repurposed fiber-optic cables that make up one of the longest land-based quantum networks in the nation.
Researchers hope to use the 52-mile quantum test site in Bolingbrook and others like it to prove that you can trap information inside a quantum state of matter (like a photon) in one location, send it somewhere else, and access it completely intact on the other end. They need to factor in the challenges of frozen ground, the sun’s radiation, and vibrations from all those vehicles traveling overhead, but if they can prove it, they’ll have invented a way of communicating that makes 5G seem quaint. Researchers at other laboratories are simultaneously trying to feed algorithms into similar elementary states of matter, known as quantum bits, and have them come out transformed correctly at the end of the computation. If that’s successful, they’ll have an entirely new type of computer on their hands.
It’s been clear to physicists for years that the long-established principles of quantum mechanics can revolutionize computing and the internet. If quantum bits can be tamed, they could run algorithms in just a few seconds that would otherwise take years to complete. Stable photons could transfer information across the world instantly in a way that likely could never be hacked while in transit, since any disturbance would destroy the information.
To the rest of us, the quantum revolution might seem as if it has just transformed from a sleepy scientific theory into the sharpest of bleeding edges. It’s even possible that we’re currently experiencing something of a quantum bubble—and that it might be about to burst. In 2017, most of the quantum test loops were just dormant fiber-optic cables, and no one had been able to get quantum bits to reliably process information in the same way classical computers can. Now, there are more than a dozen functioning quantum computers around the world, a few of which any software developer can access via familiar services: say, an Amazon Web Services account.
Within the past two years, America has committed more than $1 billion in government funds to quantum information research, quantum computing startups have closed multiple venture funding rounds, and IBM announced that it is forging ahead with plans to build a computer with more than a million quantum bits, up from a maximum of around 60 today.
Despite advances coming at a breakneck pace, many of the people working in the nascent field of quantum information science acknowledge that quantum states are not yet reliable or understood well enough to replace traditional computing and the internet. Some believe they never will be—that no one will ever buy a phone with quantum bits instead of an Apple A12 Bionic, and that quantum bits and other elementary particles will forever be relegated to scientific research.
Qubits: Way Better Than Zeroes and Ones
A computer made up of quantum bits—qubits, for short—is really a collection of circuits. As in a classical computer made up of bits, the input values proceed through a series of logic gates in the circuit, each of which modifies the value to produce an output. The most important difference between quantum computing and classical computing is that bits are binary. They are either up or down, open or closed, zero or one. Qubits, on the other hand, can be entangled—present in multiple states at once, a so-called superposition. (Watch the video above, from Rigetti Computing, for more details.)
If you’re trying to solve a complex algorithm, say, as part of a software application to run on a classical computer, you’ll need to string together multiple bits of zeros and ones. But if you’re running an algorithm using qubits, you might need only a single qubit in a superposition to replace all those classical bits. String multiple qubits together into a quantum circuit, and the possibilities are staggering. Theoretically, you could run an algorithm so complex that there’s no analog to classical computing as we know it.
The most difficult problem to solve in improving quantum computing and communications is the fragility of the quantum state of matter. We are starting to be able to protect traveling quantum particles against the effects of weather and road vibrations, but only in test loops—not over the thousands of miles required to replace the current internet. Likewise, no one has yet figured out how to make qubits function reliably, even in a controlled laboratory setting.
They work well enough in small groups and confined to specific types of computations, as IBM demonstrated using a stable 27-qubit computer called Falcon earlier this year. They’re mostly useful for testing purposes: Researchers can feed them problems with known solutions and then validate their answers. But so far, qubits have proven too fragile to function reliably in larger groups, which effectively limits their ability to graduate from beta and accurately perform any computation a classical computer would.
“As we push on the number of qubits, you’re able to explore a much more varied set of quantum circuits,” says Jerry Chow, the senior manager of the Experimental Quantum Computing Group at IBM. If only it were that simple. The “lossy qubit” problem, as Chow puts it, means that parts of each quantum computer that exists today are dedicated just to resolving errors in their computations, instead of performing the computations themselves. The quantum volume of a computer, a numerical value that describes its maximum potential to perform calculations, is always less than the number of qubits it contains. Likewise, the number of photons that begin their journey intact at the beginning of a journey through a test loop is always greater than the number that return.
To circumvent this problem and unlock the full potential of quantum computing, some researchers are working on adding error-correcting codes, which are already implemented in some classical computers. Others are exploring alternative methods of applying quantum physics to computing that don’t involve gates and circuits. One possibility is tricking quantum particles into ignoring background noise—vibrations, temperature changes, and stray electromagnetic fields, for example—that causes them to break down. A University of Chicago team announced in August that they had successfully performed this kind of trickery in a limited experiment.
Quantum annealing is another technique with potential. It involves harnessing fluctuations in quantum states to perform calculations instead of sending them through gates in a circuit. Some commercially available quantum computers from D-Wave, a small Canadian firm, use this method. But they also suffer from errors, and so far, they’ve proved effective only at solving specific types of algorithms, like those based on the “traveling salesman” problem, which seeks to find the shortest possible route between a set of points. Volkswagen used D-Wave’s approach in an experiment last year to help buses in Lisbon, Portugal, escape traffic jams. The experiment was declared a success, though it was limited to taking attendees of a technology conference from the airport to the convention center.
The most infamous example of the lossy qubit problem surfaced in October 2019, when researchers at Google announced they had completed a benchmark test on a 53-qubit quantum computer in 200 seconds. The test would have taken a classical supercomputer far longer—anywhere from a few days to 10,000 years, depending on its specifications. On the basis of the experiment, nicknamed Sycamore, Google claimed to have achieved quantum supremacy, or proof that a quantum computer can handle an algorithm faster than a classical computer can without making any mistakes. It’s something of a holy grail in the field of quantum information science, and Google CEO Sundar Pichai was quick to hail it as quantum computing’s “hello world” moment.
Soon after, though, researchers disputed whether the experiment was as significant as Google claimed, setting off a buzzworthy quarrel. For Wiliam Oliver, a physicist at MIT who studies qubits, the larger problem with quantum supremacy isn’t whether or not it exists, but when it breaks down.
“Most people in the world think [Google] achieved it,” he said of Sycamore. “But had they added a couple more qubits, then they wouldn’t have been able to do it.” Oliver thinks the benefits of quantum computing are more than just supremacy over classical computers. The real holy grail, he says, is for quantum computing “to be able to run anything for any amount of time without error.”
Even a year later, Jerry Chow still thinks of the Google announcement as a footnote on the journey to create quantum computers that researchers and even regular people can actually use without worrying about their accuracy or stability. “That was an interesting academic work, to push that type of problem,” Chow says of Sycamore.
Show Me the VC and Government Money
If there is a quantum bubble, it’s inflated both by the new flurry of Sycamore-type academic work and a simultaneous push from private corporations to develop real-world quantum applications, like avoiding traffic jams, as a form of competitive advantage. We’ve known about the advantages that quantum physics can offer computing since at least the 1980s, when Argonne physicist Paul Benioff described the first quantum mechanical model of a computer. But the allure of the technology seems to have just now bitten enterprising businesspeople from the tiniest of startups to the largest of conglomerates.
“My personal opinion is there’s never been a more exciting time to be in quantum,” says William Hurley. Strangeworks, the startup he founded in 2018, serves as a sort of community hub for developers working on quantum algorithms. Hurley, a software systems analyst who has worked for both Apple and IBM, says that more than 10,000 developers have signed up to submit their algorithms and collaborate with others. Among the collaborators—Austin-based Strangeworks refers to them as “friends and allies”—is Bay Area startup Rigetti Computing, which supplies one of the three computers that Amazon Web Services customers can access to test out their quantum algorithms. That service, called Amazon Braket, made its debut in August and counts Volkswagen and Fidelity Investments among its customers.
Quantum information tech is so appealing that conglomerates are now carving out entire research divisions to explore it as a way to stay competitive. The JPMorgan Chase bank has researchers developing quantum algorithms for every arm of its business, from encryption and security to options trading.
“We’re completely in a research mode right now,” says Rob Matles, the director of JPMorgan’s Future Lab for Applied Research and Engineering. “We want to be ready when quantum supremacy is met.” Matles is optimistic in particular about how quantum computing can improve options trading, an area of finance in which speed and accuracy is critically important.
All this activity is both supported and incentivized by the promise of taxpayer funding in the US and abroad. Established in 2018, America’s National Quantum Initiative is expansive in scope (it calls for a 10-year plan to “accelerate the development of quantum information science and technology applications”) and generosity ($1 billion has been authorized so far). There’s also plenty of support from the military: The Defense Advanced Research Projects Agency (DARPA) has granted nearly $20 million so far this year to spur the development of quantum computers, with no requirement that they have an advantage over classical ones.
Rigetti, which claims to have the only dedicated quantum integrated circuit foundry in the US, is a magnet for government funding and venture capital. It secured $9 million from DARPA in March, then closed a $79 million series C, and in August announced plans to build its second quantum computer in the UK as part of a consortium funded by a £10 million (around $13 million) grant from the British government.
Governments and militaries are particularly interested in building a quantum internet, and they have a special affinity for the quantum test loops like the one at Argonne. “We now have the blueprint to make this quantum internet a reality,” Department of Energy Undersecretary for Science Paul Dabbar announced in July. Eventually, the department plans to build quantum test loops at all 17 national laboratories and connect them together to create a rudimentary nationwide quantum communications network.
For all of the investment and optimism, however, there’s also a very real sense that the uncertainty of quantum’s capabilities represents a gamble. “Quantum isn’t ready to solve real-world problems,” Matles admits. JPMorgan started its quantum research about three years ago, and since then he’s watched advancements such as the Sycamore experiment with interest. But he insists that the company is still in an optimism phase and isn’t ready to speculate about the specific improvements quantum computing might offer or how long it might take to be ready.
“We’re still in the world of simulation,” Hurley says of quantum information technology. “These things aren’t computers. They’re great equipment for exploiting the quantum space, but a lot of them aren’t accessible all the time. They can’t solve anything that a regular computer can’t solve, and none of them work without a classical computer attached to it.”
Indeed, if lossy qubits are the bane of quantum physicists, then access challenges are the bane of corporate researchers. Every time a quantum computer finishes running its algorithm, it needs to rest, or else the quantum entanglements will quite literally collapse. The risk that qubits will lose their superposition and the information they can hold is known as decoherence, and it’s further evidence of the fragile nature of quantum computers.
In September, Oliver and other scientists announced that they believe otherwise harmless radiation from common objects like concrete walls hastens this decoherence. Relegating quantum computers to radiation-free bunkers is impractical, and the research into other potential remedies, such as background noise trickery, has only just begun. So quantum computers will have to be reset frequently for the foreseeable future. If you’re trying to offer quantum computing as a cloud service, as Rigetti and Amazon are, that means a lot of waiting time for your customers.
will keep qubits superchilled. (Photo credit: IBM Research)
Access challenges are compounded by the fact that you need a way to communicate your algorithm from the classical computer to the quantum one. Originally, Rigetti allowed customers to submit a circuit over the regular internet and get the results back later. “As a way to evaluate how to construct a circuit, it works just fine,” explains David Rivas, Rigetti’s senior vice president of systems and services. “But what you really want is a tight loop between the classical computer used to submit the results and then having the circuit run again. Sticking the public internet in the middle of that is a significant deterrent.”
Rivas says that Amazon customers will be able to avoid some of this lag time thanks to recent improvements, which allow Rigetti to evaluate thousands of circuits simultaneously and return the results to the customer’s classical computer “on the order of milliseconds.” But he acknowledges that short of a full switchover to quantum, the only way to avoid these lags completely is to integrate quantum and classical systems, an achievement he thinks is decades away. It doesn’t take a doctorate in physics to see that the prospect of two computers in different cities or on different continents communicating exclusively via quantum states will take decades. That’s if it’s physically possible at all.
Will Quantum Computing Ever Matter Outside the Lab?
Judging from the standpoint of the past three years of breakneck quantum advances, waiting decades for the next great leap in quantum information technology seems like an eternity. But progress is relative. More than a century passed between the first electronic circuits of the 1900s powered by vacuum tubes and the advanced semiconductor-fabrication techniques that make it possible for a postage-stamp-sized A12 Bionic to power an Apple iPhone. And the internet existed as a rudimentary Department of Defense project decades before AOL. So quantum pioneers are eager to pontificate about what the future might hold.
David Aschwalom is a physicist at the University of Chicago, the director of Argonne’s quantum test loop project, and the lead author of the background noise trickery study mentioned earlier. He figures that the current state of quantum information research is roughly equivalent to 1950’s-era classical computers with a few dozen transistors (modern laptop computers have billions of transistors). But he points out that the equivalent quantum machines, those with a few dozen qubits, “scale in a way that’s highly non-linear.” By the time someone invents a quantum computer with around 200 qubits, we’ll be able to process algorithms with more states than there are atoms in the observable universe, Aschwalom says.
For academic researchers, that’s as powerful a motivator as a competitive advantage is to the likes of JPMorgan. And there are several companies motivated by both scientific advances and profits. IBM in September published a roadmap of how it will get from this year’s 27-qubit computer to a 1,121-qubit processor named Condor by 2023. Eventually, the company hopes to build a fully fault-tolerant computer composed of a million qubits, though it did not peg a target year for that achievement. It’s a daunting project, even to IBM’s own engineers. Of a functioning quantum computer of that scale, Chow says, “I can draw on a piece of paper what it might look like, but I’d probably be wrong.”
Aside from the challenges of getting that many qubits to play nicely together and maintain their coherence, Condor will also require advancements in support systems and physical architecture. A quantum computer with low-temperature superconducting qubits is a behemoth and a curiously beautiful sight to behold, with multiple dilution bridges and cryogenic cooling chambers. Tubes and hoses connect all the parts together.
All computers generate heat, but quantum computers are veritable furnaces—and they’re extremely susceptible to radiation and temperature fluctuations in the first place. IBM points out that today’s commercial refrigerators will not be capable of effectively cooling and isolating a million-qubit computer. So new refrigerators must be on the quantum roadmap as well. One potential design is a 10-foot-tall and 6-foot-wide super-fridge.
“Ultimately, we envision a future where quantum interconnects link dilution refrigerators each holding a million qubits like the intranet links supercomputing processors,” IBM Quantum Vice President Jay Gambetta wrote in a blog describing the project.
Visions of a Quantum-Powered Future
The research is very real, but such visions border on fantasy, or at the least are difficult for most of us to conjure. All of the algorithms that the 10,000 software developers share on the Strangeworks platforms are simply suggestions of what might be. To validate them, they must first find spare quantum computing time and then be verified for accuracy after they’re run. Developers could instead simply rewrite those algorithms to run on classical computers and call it a day.
Even if a quantum computer succeeds at completing them faster than a classical one, there isn’t agreement among quantum pioneers that such supremacy matters much in the long run. As for the quantum internet, the disused fiber-optic cables that now serve as its testbed are stark reminders of the challenges we still haven’t solved in the effort to bring today’s broadband internet to every American household. This quantum state of affairs seems like it’s ripe for setbacks, at best. At worst, it’s a bubble with the potential to disappoint both taxpayers and venture capitalists.
But from a consumer’s perspective, quantum-based technology is not like 3D televisions or the Oculus Rift. Quantum’s greatest potential doesn’t require you to buy a new TV or dedicate an entire room of your house to video games. But even if it turns out to be impossible to make phones with quantum processors that don’t seize up when they get hot, or to send information encapsulated in photons across continents reliably, the general public will likely still feel quantum’s contributions to everyday life. Solving the coherence problem will require unknown feats of ingenuity, but it’s easy to see how faster buses and options contracts will make commuters and investors happy, even if the computations are performed in an isolated bunker and then delivered over the regular old internet to a Pentium-powered tablet PC in a rattling old bus.
“The average consumer is likely to see quantum computing in their lives in one of two ways,” Rivas believes. The first is in improvements to the traveling-salesman problem. Whether that’s redirecting Uber drivers to city neighborhoods with high demand or navigating buses around traffic jams, improving the ability to get between multiple locations as quickly as possible is a huge boon to modern life.
The second way that quantum could have an impact on everyday life without completely taking over computing and the internet, Rivas says, is in improvements to consumable goods, especially drugs: Quantum computers have proven adept at building molecular models used in drug development. Both of these cases and others like them suggest a future in which quantum physics does not replace the current information technology infrastructure we have today. Really, the future of quantum is the future of the data center, not what phones or laptops will look like in 50 years. “Our interest is building the replacement to the supercomputer,” Rivas says.
Put that way, perhaps there is no quantum bubble about to burst. The importance of supercomputers is easy to understand today, even if it’s not always easy to describe exactly what they do. Improving these types of machines with quantum tech seems easier to justify, whether or not we ever reach quantum supremacy or invent indestructible photons. Quantum information’s future could be seen as an analog to Manifest Destiny, the view that the westward expansion that swept 19th-century America was both defensible and inevitable.
“It used to take months to travel from New York to California,” Hurley points out. “And then came the train. Think of trains as classical computers.” More than 150 years after the Golden Spike marked the completion of the transcontinental railroad, hardly anyone marvels at trains. “But no matter how far you go, you eventually reach an ocean,” Hurley says. “So you need air travel.”
That’s quantum computing: a revolutionary technology that could allow us to cross metaphorical oceans. Even in maturity, it might still rely on classical computers to perform common tasks, just as the denizens of the suburbs take airplanes to go on foreign vacations but arrive at the supermarket and the mall in SUVs on 15-lane roads. If the recent pace of money pouring into quantum computing research has you itching to buy a quantum-powered iPhone, you’re probably living in a bubble. But the quantum breakthroughs of the last few years suggest changes in the not-too-distant future that are both less obvious and more revolutionary. New vaccines could be born in days instead of years, and stock trades will settle almost instantly, which makes it worthwhile to spend a few billion dollars to speed up photons and build giant refrigerators today.
keyword: A Bubble Ready to Burst?A Bubble Ready to Burst?A Bubble Ready to Burst?