Superpowerful computers will extract enormous benefits from artificial intelligence—and deliver proportional advantages.
Early in October, US President Joe Biden traveled to Poughkeepsie, New York, and stood shoulder to shoulder with IBM executives as they announced that the company would be spending $20 billion on production of semiconductors and development of advanced technology like artificial intelligence (AI) and quantum computing (QC) research.
Both AI and QC will be critical for corporate effectiveness and national security in coming years. Linking AI and QC in the same news cycle was understandable too, given that QC represents to some degree the logical extension of AI.
AI is often used in automating tedious tasks. Many of its achievements, such as facial recognition, are simply the result of raw computing power. Quantum tech expands computing power exponentially. Some believe it will eventually help doctors identify cancers earlier, pharmaceutical companies bring drugs to market faster, extend the life of batteries and more.
It may not take decades before real-world solutions emerge, either. “If publicly available vendor road maps hold, it is entirely possible to have quantum-enabled AI use cases in five years,” says Scott Buchholz, emerging technology research director and CTO for Government and Public Services at Deloitte Consulting.
Indeed, some believe corporate and government leaders should begin laying out quantum-computing strategies now, “especially in industries, such as pharmaceuticals, that may reap the early benefits of commercial quantum computing,” says consulting firm McKinsey. While most activity is still in research or pilot phase, a few startups like Paris-based QC firm Pasqal are getting close to rollout. Pasqal already has a customer list that includes Johnson & Johnson, LG Electronics, Airbus, BMW, EDF, Thales, MBDA and Credit Agricole CIB.
Quantum computing is based on “qubits” (quantum bits), which are the smallest units of quantum information, analogous to the regular computer bit. Qubits, IBM explains, “harness quantum mechanical phenomena … to solve problems that are fundamentally intractable for classical computers.”
The sheer power of QC is mind boggling. A 2017 article in Scientific American claimed that “in principle, a 300-qubit quantum computer could perform more calculations at once than there are atoms in the observable universe.” That statement is now oft-repeated industry lore. Pasqal’s processors have already reached 100 qubits—close to the 127 qubits of IBM’s ground-breaking Eagle—and Pasqal expects to bring a 1,000-qubit processor to market by the end of 2023. Meanwhile, companies like IBM, Microsoft, Google and Honeywell have been investing heavily to expand the technology. In November, IBM unveiled its 433-qubit quantum processor called “Osprey.”
“Quantum computing has the potential to ultimately be game changing in a number of ways,” says Charles Toups, vice president and general manager for Boeing Disruptive Computing and Networks. It could enable the aerospace giant to model complex materials more accurately and also “faster than we are able to complete small-scale, lower fidelity models today,” he says. Boeing sees promise, too, in using QC to model detailed chemical reactions “that will help us devise longer lasting protections for corrosion and ultraviolet exposure.”
The first areas likely to receive a QC boost are optimization, machine learning (ML—an essential component of AI, the terms are often used interchangeably) and predictive modeling, where finding solutions is “fiendishly difficult” but can be easier with quantum computers, given their “ability to handle data that has high degrees of ‘optionality,’” or dimensionality, Buchholz adds
“This is the quantum decade,” says Konstantinos Karagiannis, director, Quantum Computing Services at Silicon Valley consultant Protiviti. “The uses aren’t as far away as most people think.”
One type of proto-quantum computer already in use is the “quantum annealer.” Annealers work hand in hand with classical computers to solve optimization problems like routing trucks or assembling factories. But this is just a taste of things to come, especially when gate-based “universal” QC machines pioneered by IBM, IonQ and others reach the market. “Those are the ones that have the potential one day to crack encryption and do some of those bad things,” Karagiannis adds, “but also to change the world, because quantum machine learning will run better on them.”
Protiviti is now employing a gate-based quantum computer for fraud detection, a typical ML classification use case. “There are only a handful of machines on the cloud that are powerful enough to do it. You have to wait for your turn to run it,” says Karagiannis.
Possible Eco Benefits
The field appears to be well funded. Venture capital firms alone invested more than $1 billion into the sector in 2021, according to Deloitte, which says the industries likely to gain first from the technology include pharmaceuticals, chemicals (catalysts, more eco-friendly feedstocks), automotive (optimized factory-robot motions) and finance (investment portfolio optimization).
QC offers potential ecological benefits too: Quantum computers using “quality” qubits (still not commonly available) could make high-level computing dramatically more energy efficient, according to Interesting Engineering. “There are tasks for which the quantum computer could spend one hundred times less energy than the best current supercomputers.”
Quantum-enabled AI also has a national security aspect. Some believe China has gotten the jump on the US and other Western nations in the AI race, and Bloomberg reported in mid-October that the US may soon place export controls on AI and QC technologies. Others worry that QC could one day break the algorithm used for the public-key encryption that allows secure transmission of email and much of modern communications.
While use cases involving universal QC are still rare, more companies announced AI/QC initiatives in the past year. LG Electronics is exploring QC for big data, IoT, AI, robotics and other applications that, LG notes, all require processing a large amount of data. Meanwhile, HSBC is also looking at QC for investment portfolio optimization, risk mitigation and fraud detection, which typically use ML algorithms to process large amounts of data.
This is the first technology that is truly exponential, says Karagiannis. Many people today are familiar with Moore’s Law, which roughly states that computer power doubles every two years. But Moore’s Law pales beside QC’s projected growth curve, where a system’s power doubles every time a single qubit is added. “[If] someone has a 100-qubit machine and then someone else has a 200-qubit machine, that’s many, many orders of magnitude more powerful, not just double,” Karagiannis explains.
Building on the success of its 127-qubit Eagle, IBM in 2023 will deliver the Condor, “the world’s first universal quantum processor with over 1,000 qubits,” according to the company, and a 4,000+ qubit processor by 2025.
Not surprisingly, obstacles remain before these science fiction-like capabilities can be realized. A quantum computer is not just a scaled-up version of today’s supercomputers. “It’s something very different,” says Deloitte’s Buchholz. “We are attempting to constrain individual atoms and particles to behave in a regimented fashion, which is not their natural state.” They need to be interconnected, contained and constrained, something “unthinkable” a decade ago.
“Some approaches to quantum machine learning introduce completely novel ways to do machine learning that do not have a classical analogue,” says Vedran Dunjko, associate professor at the Netherlands’ Leiden University. But many of these “novel” uses are many years away—and may never be realized. In terms of first applications, Dunjko anticipates use cases close to quantum technologies, as in the development of exotic materials.
German conglomerate Bosch also believes that quantum computers may soon offer “a significant advantage over conventional computers in discovering and designing new materials” for products like fuel cells, batteries, electric engines or advanced sensors. “Classical computers are not able to calculate the properties of these materials with sufficient accuracy,” the company commented recently in announcing its new partnership with IBM to develop quantum algorithms for industrial applications.
Aerospace firms are watching developments closely too. “Harnessing quantum technologies for the aerospace industry is one of the great challenges we face in the coming years,” commented Greg Hyslop, Boeing’s chief engineer and executive vice president of engineering, testing and technology, in discussing Boeing’s $5 million gift for faculty research in quantum science.
“The emergence of some small-scale prototypes of QC has already propelled the development of post-quantum cryptography,” says Takaya Miyano, a professor of mechanical engineering at Japan’s Ritsumeikan University. He expects to see the first practical uses of QC for cryptography and the military as well as pharmaceuticals.
The US-China Rivalry
“AI and quantum computing are among the most important technologies for the next decade and beyond,” Jonathan Panikoff, senior fellow at the Atlantic Council’s Geoeconomics Center and director of its Scowcroft Middle East Security Initiative, tells Global Finance. AI and QC are also “very likely to underlie portions of US-China competition for the coming decades as well.”
Admittedly, it may take time before this all plays out. Devices like China’s powerful 66-qubit Zuchongzhi processor “are not capable of universal quantum computation and cannot play a role in such geopolitical struggles,” Marek Narozniak, a physicist and member of a quantum research group at New York University, tells Global Finance.
Still, in one study, Chinese researchers found that Zuchongzhi was able to complete a sampling task in 1.2 hours that they estimated would have required a 2019 classical supercomputer 8.2 years to complete. But even study co-author Chao-Yang Lu acknowledged that the sample task completed had no practical value, adding that “the computational problems that can truly benefit from quantum computing are still quite limited,” as quoted in Physics World.
Governments are paying attention too. In May, a US National Security Memorandum flagged the potential “risks of quantum computers to the nation’s cyber, economic and national security,” and directed US agencies to begin a “multiyear process of migrating vulnerable computer systems to quantum-resistant cryptography.”
That White House memo “accurately captured the challenges quantum computing could pose to Western interests across the economy, finance, infrastructure and national security sectors,” Panikoff comments, while adding that “quantum computing and its uses, including for nefarious purposes, is still in its infancy and developing.”
Much Fanfare, Little Utility
What about critics like Oxford University physicist Nikita Gourianov, who recently wrote in the Financial Times that “the quantum computing industry has yet to demonstrate any practical utility, despite the fanfare”?
“There have not yet been any clear demonstrations of practical utility,” answers Boeing’s Toups. The critics have a point. “However, based on our investigations, we see great promise to eventually be able to use quantum computers to make practical improvements that will dramatically improve our products and services.”
“Those quotes always baffle me,” comments Karagiannis “There are about 140 companies that are currently using D-Wave [quantum annealers] in real business applications daily.”
In a test run, Volkswagen used live access to a D-Wave quantum processor to optimize the painting process in one of its factories.
“For example,” a case report by D-Wave Systems explains, “if a small subset of minivans was slated to be painted black rather than white, the algorithm would specifically assign those paint jobs to minivans falling within stretches of the production run where other vehicles are already being painted black.”
Volkswagen later reported, “We managed to reduce the color switches in the entire sequence significantly.”
Still, as noted, D-Wave machines, which debuted in 2011, are basically analog computers, “meaning they only tackle one specific problem,” Samuel Mugel, CTO at Multiverse Computing, a quantum software development company, tells Global Finance. “They are not equivalent to universal quantum computers, which can tackle any problem.” And regarding the latter, the technology isn’t ready for mainstream industrial use, as even Karagiannis acknowledges.
“Right now, we’re in the ‘noisy’ intermediate-scale quantum era,” Karagiannis says. “The qubits interfere with each other. The support circuitry interferes. It’s really hard … It’s hard to get these machines to stay in a behaving state.”
But the buzz is rising. “It’s almost inevitable that there will be under- and overselling as we transition from research to engineering to application,” adds Deloitte’s Buchholz.
“It is easy to be a naysayer and negative these days,” says Dunjko. He prefers to focus on how far QC has evolved since he began working in this area in 2010. “Even if technology continues just linearly, we will have tremendous devices.”
Technology does not advance linearly, of course. Progress is more a stop-and-start affair. “Creative new ideas cause disruptive jumps,” Dunjko adds, so when projecting the long-term impact of quantum-enabled AI, some intellectual humility may be in order.
“In the same way that it was impossible to understand the impact of digitalization in the 1960s or the internet in the 1990s, we have only the barest inklings today of what might be possible,” Buchholz says.