kras99 - stock.adobe.com
Quantum computing is one of the shiniest new developments in IT. And now, it is starting to gain traction -- but the extent to which it filters into the world of data centers has yet to be seen.
Money is flowing into quantum and its role is emerging. Last November, IDC published its forecast for the worldwide quantum computing market that projects customer spending for quantum computing will grow from $412 million in 2020 to $8.6 billion in 2027. Their thesis is that continued breakthroughs will drive performance and lead to wider adoption.
Computing power drives quantum's appeal
Superposition and entanglement are concepts that drive quantum computing's appeal because they increase potential computing power, in contrast to the way that on/off or one/zero states define classical computing.
Fred Chong, chief scientist for quantum software at ColdQuanta, said that there is an exponential increase of computing power with every qubit added to the computation that exploits these properties.
"They allow n qubits to simultaneously represent 2n numbers; digital computers can only represent one of those numbers at a time," he said.
Quantum processing units could be very good at simulating physics and chemistry, optimizing problems such as logistics, or even certain kinds of machine learning. Jason Larkin, a researcher at the Software Engineering Institute of Carnegie Mellon University (CMU), said organizations are actively experimenting to find potential applications of the technology.
"IBM is already installing quantum computers," Larkin said. "For example, they partnered with Cleveland Clinic where they are planning to combine quantum computing with quantum sensing technologies."
And while acknowledging the complexity and challenges inherent in quantum, Larkin said there is a unique power of the technology for applications like probing the electron structure of molecules.
"It is also supposed to be a domain where things can be done in real time that take exponential amounts of time in classical computers," he said.
Quantum computing might need more time to develop
Despite IDC's predictions and Larkin's enthusiasm, others are skeptical and do not see any immediate future for quantum computing in typical data centers.
Franz Franchetti, an electrical and computer engineering professor at CMU, draws a hard distinction between "real" quantum computing and the devices being marketed or close to market.
"They are adiabatic or noisy intermediate-scale quantum and not real, scalable quantum computers," Franchetti said. "I would assume that such experimental devices will show up at cloud providers but more as experimental systems to show how good the provider is."
Likewise, Franchetti believes it is too early for a clear architecture to emerge, though he admits that the software layer is now "reasonably standardized." However, with current quantum computing devices, their applications are limited.
Mark Acton, a data center consultant based in the U.K., said hosting quantum computers may be a difficult hurdle for enterprises in the short term. The future of quantum computing will depend on how the technology evolves and how easy it becomes to host in a more standard data center. The supercooling requirements for quantum compute mean that traditional data centers are not set up to host this equipment yet, he notes.
"The only place that quantum compute is being based currently is in purpose-built areas that are effectively research facilities," Acton said. "Quantum computing in its current and near-term future iteration will augment digital computing by being extremely fast and extremely efficient at some types of calculations and predictions, but will almost certainly not replace digital compute for more mundane applications and simple transactions."
The commercial data centers that are being built today will be around for 20 to 30 years and they are not currently being designed to host quantum computing, according to Acton. Instead, it is more likely that quantum will go into specially designed or significantly refurbished data centers, but that is largely dependent on the degree to which the developing architectures require cryogenic cooling and other special equipment for operations.
"There are competing quantum technologies and architectures," Acton said. "Until we have a more standardized approach and consistent delivery model, quantum compute is likely to be based in bespoke data centers," which are purpose built and unique to quantum requirements rather than traditional commercial data centers.
Vendors currently offering commercial products or quantum applications include Artiq, Sinara, Zapata and IBM. Some of the other organizations dabbling in quantum, according to IDC, include startup IQM, quantum hardware startup Atos, Pasqal and Nvidia's cuQuantum Appliance and cuQuantum software development kit. Outside of the U.S., the European High-Performance Computing Joint Undertaking is funding the High-Performance Computer and Quantum Simulator hybrid project.