Earlier this week, International Business Machines said that it had deployed an IBM Quantum System Two at a research center in Japan, marking the first time such a device had left the U.S.
It was a momentous occasion for the company, which has tweaked its business strategy numerous times since its founding in 1911. This doggedness has helped IBM grow into a quantum heavyweight, culminating in the development of the Heron processor, its best-performing quantum chip to date.
“IBM has been remarkably consistent in hitting its road map goals for building a fully functioning fault-tolerant quantum computer by 2029, and this is another step,” Mark Horvath, an analyst at research firm Gartner, told Barron’s.
Quantum computers use subatomic particles like photons and electrons to transmit information. To date, no company has been able to create a fault-tolerant quantum machine, or one that can perform accurate computations even in the presence of errors. IBM’s coming Starling system is designed to catch and correct these mistakes as they occur.
Horvath pushed back on IBM’s assertion that the Heron chip powering the IBM Quantum System Two was “the most performant quantum processor in the world.” There’s no objective way to test such claims, Horvath said. However, the processor has improved circuit depth as well as better error rates than previous generations.
The IBM System Two will be installed in the RIKEN Center for Computational Science alongside Fugaku, a classical supercomputer, in what’s commonly referred to as hybrid model. This approach will allow researchers to advance research on fundamental chemistry problems and other algorithms, IBM said.
The company is moving toward a model that blends high performance computing (HPC) with quantum computing, “which is generally the way utility quantum computing is going,” Horvath said. As quantum has an advantage over classical devices on certain problems like optimization or graph coloring, the technologies can work together to solve problems.
“The classical HPC system will work on most of the problem, farming out parts of the work to the quantum computer where it has a chance at a better-than-classical answer, then combining that back with the ongoing HPC system,” Horvath explained. “This is the current way quantum computers are used in most cases.”
Quantum computers don’t consistently show an advantage over classical devices, Horvath conceded. However, “very compelling developments in noise reduction and error correction” over the past five years have helped machines return accurate results more consistently.
The analyst believes these improvements will bring quantum computers into a more widespread use over the next few years. He likened the progression to artificial intelligence, which “moved very rapidly from a niche technology into a dependable technology in a similar amount of time.”
Like other companies, IBM’s goal is to build increasingly bigger and more precise machines with the hopes this will lead to wide-scale deployment of the technology. Speaking to Barron’s earlier this month, Jay Gambetta, vice president of IBM Quantum, laid out the company’s vision for quantum dominance.
“We’ve always focused on what we need to learn to be able to scale,” Gambetta said. “Our secret weapon is that we make a new device every 17 days. You focus on how you can increase the rate at which you can learn, and you get very disciplined at that.”
IBM hopes to pioneer a fault-tolerant device before the end of the decade. Gambetta is sure the company can meet its target. “We always had a plan. I finally feel like we have an execution,” the scientist said. “I’m excited to bring this quantum computer into the world.
