The race to enterprise-scale quantum computers is on

The more I get into quantum computing, the more convinced I am that enterprise-scale quantum computers will happen.

Even though I’ve spent my whole career in emerging technologies, this statement surprises and thrills me. Not so long ago, everyone thought quantum computing was sci-fi — a physical impossibility. The mere existence of the field’s current race for scale is a fantastic achievement.

Not everyone agrees with me that it’s a winnable contest though, either in a reasonable timeframe or at all. Given the scale of the challenge to get from where we are today to large-scale useful quantum computers, am I right to believe we will get there?

Why quantum computing matters

Before I dive into the scale challenge, I want to explain why I think quantum computing is important.

The idea of a new class of super-fast computer that drew on the science of quantum mechanics was first mooted in the 1980s. The science is super cool but super mind-blowing too. Scientists envisioned a quantum computer that would solve problems using “quantum bits” rather than the binary digits (1s and 0s) relied on by conventional computers. In very simplified terms, quantum bits, or qubits, should be able to hold multiple values at the same time, allowing quantum computers to work exponentially faster than conventional computers (known as “classical” computers).?

Right from the start, quantum computers weren’t entirely about replacing their classical cousins. Instead, they were (and are) intended to tackle specific kinds of problems that classical computers either can’t solve in a useful timeframe or will never be able to solve at all. These are typically complex problems involving many factors and possible combinations.

When you start to think about all the data we can access to make the modern world work better, these sorts of problems crop up all over the place. For instance, solving them could help us simulate new molecules to speed up drug discovery, plan sustainable urban environments and optimize global shipping routes to be green as well as profitable. Many other proposed use cases are just as significant, and I’m sure there will be plenty more to come.

Quantum computing’s scale challenge

Realizing such aspirational benefits is no mean feat. So far, scientists, technologists and engineers have spent 40 years working out how to build a quantum computer in the real world. After decades of graft, the past few years have seen incredible progress.

The first major milestone was when Google achieved “quantum supremacy” in 2019.[1] The company showed that its Sycamore quantum processor could solve a theoretical problem faster than any classical computer, which was an industry-recognized step towards practically useful quantum computers. The next key milestone will be when a quantum computer can do the same for a problem with a practical real-world or business application, known as “quantum advantage”. We aren’t there yet.

Even if we get there soon, the challenge of scaling up quantum systems remains immense. The biggest hurdle is making them reliable, which gets harder every time you add more qubits to add more computing power. And you need a lot of qubits to operate at commercial scale. One thousand qubits is generally recognized as the minimum to solve useful but limited real-world problems; one million qubit machines could start a computing revolution. Today, we’re at 100+ qubits.?

As I said earlier though, the pace of progress is accelerating. Google, IBM and others believe 1,000 qubit systems are only a few years away. Both tech giants have also released public roadmaps that include reaching one million qubits — by 2029 for Google,[2] and 2030 for IBM.[3]

The three obstacles everyone is trying to overcome are noise, errors and scaling up the physical connection required between qubits. Noise refers to anything that interferes with qubits’ ability to maintain their delicate quantum state, from stray atoms to vibration or temperature changes. Such interference causes qubits to “decohere” or lose the quantum properties that give them their computing superpower. The challenge comes because qubits are incredibly sensitive to noise of any kind — even interacting with qubits to run an algorithm or read the results of one can scramble the answer. Noise, therefore, creates errors, and the error rate only gets worse as you add more qubits or need them to maintain coherence for longer.

Of course, classical computers aren’t 100% error-free either, which is why the standard threshold for a “fault-tolerant” computer is 99% error-free. However, correcting errors for classical computers is fairly straightforward, requiring extra storage and logic which are both easily accessible.[4] By contrast, quantum error correction schemes use so many qubits that there aren’t enough left for useful computation, and adding more qubits is not a simple task.

To make a hard challenge harder still, for qubits to be in the quantum state that gives quantum computers their superpower, all of them need to be physically connected to each other within a quantum computer. Achieving this is an enormous challenge that only gets more difficult as you add more qubits. As a result, it is limiting scalability of current devices and the types of problems we can ask them to solve.

Pushing quantum technology forward

Recent breakthroughs addressing these challenges have electrified the quantum community and galvanized the race to commercial scale. Most notably, in January 2022, three research teams published separate papers showing that large-scale, fault-tolerant quantum computers could be “made and used in the real world”.[5] The teams from Australia, Japan and the Netherlands not only show it is possible to make 99.9% error-free quantum computers, far above the 99% threshold, but to do so in a way that is compatible with existing technology for manufacturing silicon-based processors at scale.

The news is a particular boost for developers focused on silicon-based quantum technologies. The most prominent are the superconducting systems being developed by giants such as IBM and Google, as well as by quantum startups such as Rigetti Computing. IBM’s release of a 127-qubit processor in late 2021 was another piece of good news for this branch of the sector, along with the company’s expected launch of a 1,000-qubit successor this year.[6] Rigetti Computing’s roadmap is also promising, with near-term milestones of 1,000 qubits by 2025, and 4,000 qubits by 2027.[7]?

Developers of competing “gate-based” technologies, including photonics, trapped ions, cold atoms and quantum dots, are also working hard to find answers to the noise and errors at the core of the scale challenge. At the same time, some companies are taking a different tack and looking at what quantum computing can do now.

Quantum annealing, for example, might not be as broadly useful as gate-based technologies, but research by GlaxoSmithKline in 2021 found that D-Wave’s quantum annealing processor can already compete with classical computers in terms of solving realistic optimization problems — problems about finding the best possible solution.[8] ?

Plenty of business problems are optimization problems. Scheduling logistics, managing investment portfolios and making product design or manufacturing decisions are just three. Incidentally, despite decades of commitment to quantum annealing, in 2021 D-Wave announced plans to build a gate-based or universal quantum computer in the near future as well, [9] suggesting that annealing technology alone is not enough to secure a company’s place in the quantum computing era.

Another research area I’m watching closely is hybrid classical and quantum computing. By pairing a classical computer processor with a quantum co-processor, you can, in theory at least, access quantum’s capabilities in a more reliable way. In fact, some believe that such combinations will deliver the first practical uses of current “noisy” quantum computers (NISQ - Noisy Intermediate-Scale Quantum), while we wait for large-scale fault-tolerant quantum computers to be available.

One of my own fields of expertise, machine learning and artificial intelligence, is an ideal candidate for hybrid combinations, as it uses large datasets to tackle complex problems. I’m confident that quantum-charged AI systems could speed up solution-finding for problems as diverse as fraud detection and advanced weather modeling.

Others share my confidence (but not everyone)

My optimism that the recent technology advances, publicized roadmaps and viable use cases will lead to commercial-scale quantum computing is shared by others. But it’s not universal.

Sharing my bullish corner are my EY colleagues who are trialling quantum secured communication services developed by Toshiba and partners to secure data transmitted between two of our offices in London.[10] ?Quantum software company Zapata is also close by. They believe that tech progress, levels of business investment and tentative steps out of research and development will lead to hybrid classical-quantum systems in “real customer use” within two years.[11] According to the UK’s Financial Times, even tech giant IBM’s head of research said recently that quantum advantage within two years is possible.[12]

Google is more circumspect, however. They believe it could take up to a decade to achieve a useful, error-corrected quantum computer.[13] And although 170 organizations are already using IBM’s 27 quantum systems, it is still for research rather than operational purposes.[14]

A fair number of other experts, including within the field, believe overcoming the noise problem at scale is such a big challenge that general applications, or “broad quantum advantage” could still be decades away if we get there at all.

Levels of investment are cautiously encouraging

Investment often offers a useful indicator of attitudes to an emerging technology. For quantum computing, it tells a story of growing belief tempered with caution. On the positive side, both public and private investment have risen in recent years as the technology has advanced and the path ahead become clearer. Venture capital (VC) investment nearly tripled in 2020 before rising to over $1b in the first three quarters of 2021.[15] Quantum startup valuations are starting to heat up too. For instance, PsiQuantum was valued at an impressive $3.15b in July 2021.

Global governments including the US, China and EU have also been investing over the past decade, leading some to dub quantum computing the subatomic arms race. Military and intelligence applications are naturally high on the wish list, with quantum computing named as a top five technology in the US quest to maintain security supremacy.[16]

Looking more broadly though, investment in quantum computing is still far below other emerging technologies such as artificial intelligence ($77.5b total investment in 2021, more than double 2020,[17]) and cryptocurrency and blockchain ($33b in startup VC funding in 2021,[18]). For quantum hardware and ecosystems to reach similar levels of maturity, the sector will need a hefty increase in capital investment compared to today.

Quantum computing is coming

It is quite clear that quantum computing is a tough nut to crack. But recent technical achievements show that efforts to do so are paying off. Decoherence remains a thorny problem, but progress on error correction shows solutions are possible.

Whether it takes a few years or longer, with continued investment and ingenuity, I believe large-scale quantum problem-solving will come. In the meantime, I see innovations such as cloud access to hybrid quantum-classical systems opening up an interim era of quantum benefit, particularly when harnessed to speed up problem-solving with machine learning and artificial intelligence.

?

The views reflected in this article are the views of the author and do not necessarily reflect the views of the global EY organization or its member firms.

[1] https://ai.googleblog.com/2019/10/quantum-supremacy-using-programmable.html (Oct, 2019)

[2] https://quantumai.google/learn/map (Accessed 2023)

[3] https://fortune.com/2020/09/15/ibm-quantum-computer-1-million-qubits-by-2030/ (Sep, 2020)

[4] https://blogs.scientificamerican.com/observations/the-problem-with-quantum-computers/ (Jun, 2019)

[5] https://www.independent.co.uk/tech/quantum-computer-study-update-science-b1996393.html (Jan, 2022)

[6] https://fortune.com/2020/09/15/ibm-quantum-computer-1-million-qubits-by-2030/ (Sep, 2020)

[7] https://siliconangle.com/2022/05/16/rigettis-quantum-computing-roadmap-gets-pushed-back-amid-supply-crunch-higher-costs/ (May, 2022)

[8] https://www.zdnet.com/article/quantum-computing-quantum-annealing-versus-gate-based-quantum-computers/ (Mar, 2021)

[9] https://techcrunch.com/2021/10/05/d-wave-plans-to-build-a-gate-model-quantum-computer/ (Oct, 2021)

[10] https://www.ey.com/en_uk/news/2022/04/bt-and-toshiba-launch-first-commercial-trial-of-quantum-secured-communication-services (Apr, 2022)

[11] https://sifted.eu/articles/companies-spending-quantum-computing/ (Jan, 2022)

[12] Quantum advantage is the next goal in the race for a new computer age | Financial Times (ft.com) (Jan, 2022)

[13] https://quantumai.google/learn/map (Accessed 2023)

[14] Quantum advantage is the next goal in the race for a new computer age | Financial Times (ft.com) (Jan, 2022)

[15] https://pitchbook.com/news/articles/quantum-computing-venture-capital-funding (Sep, 2021)

[16] https://www.american.edu/sis/centers/security-technology/quantum-computers.cfm (Nov, 2021)

[17] Report: AI investments see largest year-over-year growth in 20 years | VentureBeat (Dec, 2021)

[18] Report: VCs Invested $33B in Crypto and Blockchain Startups in 2021 - Blockworks (Jan, 2022)