The Quantum Race

Yorktown Heights

3 October 2019

Earlier this year, IBM launched the IBM Q System One, the world’s first commercially available quantum computer.

Quantum computing research has been ongoing for more than 15 years. To date, its capabilities are mostly theoretical, and knowledge of the field is contained within a small circle of academics, scientists, engineers and theorists. “But the future of quantum computing is coming much faster than we think,” explains Bob Wisnieff, chief technical officer of IBM Q, the branch of IBM Research that deals with quantum computing. “By designing the System One as an integrated system, we wanted to give people an intuitive grasp of what a quantum computer is and show that it is coming together as a technology.”

The System One thus makes the leap from a theoretical laboratory experiment to a packaged commercial product. To achieve this, IBM Q enlisted London-based architects Universal Design Studios and industrial designers Map Project Office – both founded by industrial designers Edward Barber and Jay Osgerby – to design the System One’s casing and interface. For these collaborators, this was the opportunity to set the design for a revolutionary future technology. “We wanted to create an archetype that would capture the public’s imagination,” says Map’s director Will Howe. “It’s a technology that people have never seen before and we wanted to express the future possibilities of what it can achieve.”

I meet Howe and Universal’s co-principal Jason Holley at their shared studios in Clerkenwell, London, where they talk me through their collaboration with IBM Research and their designs for the System One. “It can be a bit daunting knowing that you’re often working with Nobel Prize winners and nominees,” says Howe.

“But it makes every day a victory for us,” adds Holley.

The System One is a 3cbm glass vitrine, which highlights the distinctive hardware element required by quantum computing: the cryostat. The cryostat freezes atoms at around 10mK, a temperature colder than outer space, such that the computer can perform quantum calculations. The device hangs at the centre of the vitrine, encased in mirror-polished stainless-steel. “We wanted to give the cryostat a heightened presence. We’re drawing parallels with the priceless objects and artworks that you would see in a museum,” explains Holley, “It’s an object that you can look at, but can’t touch.” The glass panels were manufactured by Goppion, a Milanese manufacturer that has also produced glass displays for the Louvre. “When scientists talked to us about quantum computing,” adds Howe, “it was usually in terms of the acts of looking and observation, so the vitrine reflects this process.”

While many people will have access to the System One over the cloud, few will actually see the machine. IBM’s earlier quantum computing devices – which lacked a definite physical form, and instead existed as discrete components scattered around a laboratory – were made accessible to the public through the cloud-based platform IBM Q Experience in 2016. This experimental platform with 5-qubit and 16-qubit hardware (as well as simulations of 32-qubit quantum computers done on classical computers) attracted more than 120,000 users, engendered over 10m experiments, and led to the publication of some 170 research papers in the fields of machine learning, optimisation, chemistry, and quantum games. The System One is currently usable through the IBM Q Network, a cloud-based commercial service that gives corporate clients from Fortune 500 companies and research institutions access to 20-qubit quantum computers. But for visitors to IBM’s Thomas J. Watson Research Center in Yorktown, New York, where the computer is currently on display, the experience of the System One is immersive. The computer is bolted to the ground of a pitch-black room, where a light box on the casing’s ceiling illuminates the cryostat. “There is a bit of theatre to it. The machine lights the room, as opposed to the room lighting the machine,” says Howe. “Its sound is analogue and mechanical, almost steampunk-ish. When you’re in the pitch-black room on your own and all you hear is the sound, it’s mesmerising.”

IBM is among the leaders in the global race to build performing quantum computers. The technology can solve complex calculations that are beyond the scope of classical computers, and could have a major impact on cybersecurity, calculations of financial risk and quantum chemistry, as well as other fields. Contenders include heavyweights such as Google, Microsoft, Intel and the Chinese government, as well as a host of startups. In December 2018, the United States passed the National Quantum Initiative Act, which pledges to invest $1.2bn in quantum computing research over the next 10 years, joining the European Union’s pledge of $1bn in 2016.

Quantum computing was developed to deal with mathematical calculations that cannot be solved by a classical computer. The latter relies on bits to store data, which can be in states of either 1 or 0. Some problems, such as the factorisation of very large numbers, would take a classical computer longer to work out than the age of the universe. Quantum computers may bridge this gap, however, due to a process called superposition. This allows quantum bits, or qubits, to occupy the two states of 0 and 1 at the same time. To put that in perspective, 50 qubits can represent more than one quadrillion data values simultaneously. This principle exponentially increases computational speed, so that calculations that are deemed ineffective on a classical computer could be resolved in a matter of days by a quantum machine. By drawing on quantum mechanics, quantum computers use the physical phenomena of nature to manipulate information. The theory is that this will enable scientists to accurately represent natural phenomena in simulation. For example, in 2017 IBM scientists simulated beryllium hydride with a quantum computer. While qubits are not made of beryllium hydride, they can be programmed to precisely measure and simulate the molecule. The computer’s potential to represent nature with such accuracy is reflected in the visual aspects of the design of the System One. “We used pure geometrical forms [a cube, circles and rectangles] to symbolise this relationship with nature,” says Howe, citing the golden ratio as one example of the theories used to achieve the design.

But the wider impact of quantum computing on security and information processing is also a subject for debate. “The most lasting effects of reliable quantum computers are not about the quality of your YouTube videos, but rather about money and power,” says Bryan Roberts, a philosopher of physics at the London School of Economics. “In the new economy, money and power are mined from data. And, more valuable data is accessible to those with higher computational speeds. So, computational speed is a royal road to money and power.”

The most notable of these effects is the threat that quantum computers pose to encryption. In theory, powerful quantum computers could execute Shor’s algorithm, which deals with integer factorisation and underlies most global encryption methods. “Encrypted information can be recorded and stored, with the knowledge that it can be decrypted in 20 years when the technology becomes available,” says Joseph Rahamim, a PhD candidate in quantum computing at the University of Oxford. IBM is developing software in response to this threat, which would protect clients and governments from quantum computers in the future. In addition, global research into space-based quantum communication aims to guarantee security for satellite-to ground and inter-satellite communications. “The useful advantages of quantum computers – machine learning, optimisation and chemistry – will come well before the machines can be used for cryptographic work,” Wisnieff insists. “But I worry about what governments can do with quantum chemistry,” says Rahamim, highlighting its potential for the development of chemical weapons.

As such, quantum computing research is as closely linked to defence as cybersecurity. This relationship between technology and defence is a longstanding one and echoes IBM’s work during the Cold War period. The company famously developed SAGE, the national air-defence system implemented by the United States to warn of and intercept airborne attacks during the Cold War. “Between 1952 and 1955, it generated 80 per cent of IBM’s revenues from computers, and by 1958, more than 7,000 IBMers were involved in the project,” the IBM website states. “Due to the indispensable role that IBM played, alongside many other large corporations, in wartime production, its top executives had unfettered access to the highest levels of U.S. military command and to the Truman administration,” writes architectural historian John Harwood in his book The Interface: IBM and the Transformation of Corporate Design, 1945-1976. Quantum computing may well prove part of a new Cold War- style race for information, when information means power.

The cryostat, which freezes atoms at around 10mK, is critical in enabling the computer to perform quantum calculations.

The casing for the System One represents the first visualisation of a quantum computer as a piece of commercial hardware. Although the design is huge, the System One is three times smaller than current laboratory-based quantum computers at IBM, and reduces servicing times by 10, according to IBM engineers. “It’s the first version of the technology that can be produced by regular engineers with no specialism in quantum computing,” explains Rahamim.

But throughout the two-year project, the development process often brought the designers into conflict with the engineers. “There was an enormous amount of back and forth between the IBM team, Universal, Map and a wide range of physicists, engineers and scientists,” says IBM’s Wisnieff, who likens the process to “hostage negotiations”. “From an engineering standpoint, when something works, the most risk-free solution is to keep it in its place,” he says. “Even basic things, like how to route the cables became major points of the design debate.” In the first year of development, nine major test models were released during internal meetings that took place almost monthly.

The most radical change proposed at the start of the project by the designers was to remove the four-poster frame that originally supported the cryostat in the laboratory, and to use a cantilevered frame instead. “We wanted the cryostat to be the focal point of the machine and to hang as if it were floating,” says Holley. “The negotiations took almost five months,” says Wisnieff. “The engineers worried that removing the frame would cause vibrations that would affect the computer’s performance.” To everyone’s surprise, the new model performed better than the original and the cryostat was able to reach lower temperatures. “We didn’t know it would improve the performance,” concedes Howe.

“It was a critical moment in our collaboration. It worked to make everyone feel comfortable,” Wisnieff recalls. “Once we got the engineers’ commitment, it opened up discussions to all other design aspects.” One of the main challenges was creating the first ever public interface for a future technology whose functionalities are still unclear. “A quantum computer is not a consumer product like an iPhone and it won’t be anytime soon. It’s a technology the world has never seen,” says Scott Aaronson, a quantum computer theorist at the University of Texas. The focal point on the cryostat communicates what is distinct about the quantum computer in terms of engineering. “When we first went into the laboratory, we naturally gravitated to the cryostat, as it was the most unfamiliar object in the room,” says Howe, “so we knew that should be the focal point.” Holley shows me linear drawings of cubes and circles, in varying juxtapositions, which were drawn in the early stages of the design process. In their simplicity and emphasis on form, they echo constructivist paintings, but they also highlight the designers’ architectural approach to the design of the quantum computer.

Howe describes the System One in contrast to the “literal and metaphorical ‘black box’ hiding the inner workings of a classical computer”. While classical computers today have an opaque casing, the transparent vitrine echoes early iterations of the IBM mainframe computers, in which the computers’ inner mechanisms were visible through glass. “[IBM] attempted to establish a visual and spatial relationship – an aesthetic relationship – between the user and the computer by removing the opaque panels and offering the user a view (through a glass window) into the machine’s innards,” writes Harwood. But though the vitrine of the System One purports to reveal the inner workings of the machine, the cryostat’s mirror-polished stainless-steel casing reflects back onto the viewer. This language of mirroring, Howe argues, helps to visualise one of the most fundamental aspect of quantum mechanics. “It demonstrates a state in which particles can exist in two or more places at once.” But the vitrine also implies transparency in a technological race that is shrouded in corporate secrecy. “I don’t know what IBM’s qubit-chip looks like. They don’t show it to anyone!” says Howe.

As well as establishing the look of the quantum computer itself, the designers created a modular casing so that existing technology can be replaced with new versions. “It was designed with upgradeability in mind,” says Wisnieff. “The transition towards a commercial system means that it must be easy to maintain and that customers can make full use of it.” One of the main engineering challenges in quantum computing is scaling qubit chips. System One operates on a 20-qubit chip and the company projects that it will have 100-qubit chips in the next 10 years. Currently, the world’s largest qubit-chips are IBM Q’s 50-qubit chip (2017) and Google’s 72-qubit chip (2018). The System One casing was designed to adapt to 200 qubit hardware, giving it a projected lifespan of 10 to 15 years. “I’m really excited about how it will evolve as a design. This is going to progress over and over again. It might get bigger or smaller,” says Howe. “It could end up the size of a building,” adds Holley.

Yet one of the main challenges to the design was bringing all the elements of the quantum computer together while keeping them in isolation. Qubits are incredibly delicate, and any vibrations, ambient noise or temperature changes cause them to lose their properties. “Quantum computers can only run for a very short time before they start to pick up so much noise that they become useless,” explains Roberts. “The distinctive phenomenon of quantum computing, called entanglement, diminishes rapidly when the entangled system interacts with its
environment.” Engineers in quantum computing are working towards achieving longer “coherence times” – the period during which a quantum computer is able to perform its calculation before it picks up too much noise. The cryostat freezes the qubits, isolating them from any interference, and is therefore one of the key elements in the process. IBM is known to have achieved the longest coherence time to date: 132 microseconds. But to measure the overall performance, IBM scientists developed a metric known as “quantum volume”, which also takes into account different measurements such as qubits, connectivity, coherence time and gate and measurement errors.

The System One’s casing is designed to address these issues around decoherence. In its original laboratory setting, thousands of elements that make up the quantum computer are isolated from each other to reduce interference from the machinery’s ambient noise. “The machine is so sensitive that we had to make sure all the parts co-existed without touching,” explains Howe. Thus, the cryostat is attached to the raised floor using a cantilever frame; a hidden frame contains a helium pump for coolant and electronics; the vitrine isolates these frames without touching them. “These frames are bolted to the floor and dampened so that they absorb external vibrations,” says Howe, who adds that the vitrine also has an important functional role. “The vitrine had to be airtight to protect the cryostat from any external interference, but also serviceable.” The glass panels open easily, using a sliding mechanism encased at the bottom of the machine, which reduces servicing time.

The launch of the System One is mostly symbolic: quantum computers are still weak and unreliable machines that do not yet satisfactorily outperform classical computers. Though corporate clients and institutions are accessing the technology through cloud-based services, it is unlikely that they will buy a quantum computer any time soon. “Quantum computers will be available in normal machine rooms in the future, but I have no idea when that would be,” says Wisnieff. “Selling time on the cloud seems like the most probable model for now, as the machines are evolving yearly.” Decoherence continues to present challenges to the performance and reliability of quantum computers. “I placed a bet 15 years ago with my old college roommate (now a senior developer at Google) that this problem would not be overcome in the next 30 years,” says Roberts. “I’m still optimistic that I’ll win the bet.”

As such, some researchers view the release of the System One as premature. “I don’t care about the casing – I care about the performance of what’s inside!” says Aaronson. “Let’s prove that it works, and that it outperforms classical computers at something, before worrying about the packaging or the user interface.” But IBM insists that its product launch was a necessary step in the race for quantum computers. “Quantum computing really exists; it isn’t just a theoretical construct,” says Wisnieff, who views the System One as a timely expression of the advances in the field. “Today we can do algorithms of increasing complexity. By the time we get to a 100-qubit machine we will be able to do calculations that can’t possibly be done any other way. And that’s not that many years away.”

Developing the IBM Q System One.

Throughout the post-war decades, IBM pioneered the design of business machines whose opaque casings and simple geometrical shapes dominated the look of offices globally. From the 1950s through to the mid-70s, IBM’s design director Eliot Noyes enlisted designers and architects such as Charles and Ray Eames, Paul Rand and Isamu Noguchi to work on the company’s products as well as its corporate environments. In The Interface, Harwood describes how IBM’s design strategy revolutionised corporate environments through its focus on hardware interface design: the point of contact between human and machine and workspace. “The design programme at IBM[...] set standards of practice that quite literally changed the technics of corporate and architectural culture alike,” he writes. “In its own words, repeated through much of its promotional literature, IBM was a business whose business was how other businesses do business.”

But today, the company is struggling to modernise. From 2012 to 2017, it suffered from an annual erosion of revenue. Its iconic formulation of the laptop computer, the Thinkpad designed by Richard Sapper, was spun off to Lenovoin 2005. In 2014, it sold its Intel x86 server line to the same company, shortly after losing a $600m bid to supply servers to the CIA. Amazon, with its big-data centres that dwarf IBM’s internal architecture, was awarded the deal. While IBM continues to produce mainframes and servers, it has shifted part of its focus to developing software and cloud- based services while funding its research into quantum computing and AI. Last year the company invested $5bn of its own funds into this research arm.

The launch of the System One is part of a wider design strategy for IBM Q. “The aim is to showcase IBM’s innovations in research and to highlight the company’s dedication to open-source experimentation,” says Holley. Universal and Map will also be working with IBM on a public display room for the System One in Yorktown, and designing the test labs at the IBM Q Quantum Computation Center in Poughkeepsie, among other projects involving film and digital art. Under this programme, clients and visitors will be able to visit the labs on guided tours. “IBM have realised there’s great work going on inside closed doors,” says Holley. “It will be a great experience to open up these labs so you can observe people working. It’s part of their wider goal of opening up and inviting people to engage with them.”

As such, while the System One purports to be a commercial product, it actually plays a much greater role in strategic positioning. Rather than inspiring business efficiency, the System One’s interface emphasises the potential of a future technology. In its geometric abstraction and theatrical use of lighting, the interface gestures towards a dream, rather than the quantum computer’s current functionalities. “The relationship between the glass vitrine and cryostat is one of unobtainable science,” says Howe. “Our motivation now is to try and push the performance of these systems as far and as fast as we can,” adds Wisnieff. It evokes both the quest for knowledge and the absence of it.

The System One is a response to a technological race that is mired in uncertainty. “Could quantum computing be impossible for some deep reason that nobody has figured out yet? In some sense that’s the most exciting possibility,” said Aaronson at TEDx Dresden in 2017. “The idea that eventually with enough money and effort you could build a quantum computer as the theory said and give huge speed ups for certain things is the boring and conservative possibility.” The System One may be a transparent vitrine, but it sits nonetheless in a pitch-black room.