To deliver on their promise to revolutionize computing as we know it, quantum machines must produce results we can actually trust. But if a quantum computer solves a problem too complex for any traditional system to handle, how do we know that the answer is right?
IBM and its research partners set out to address this question in three technical papers on Thursday. Big Blue teamed up with the University of Chicago, longtime collaborator Riken, and a trio of software developers -- Qedma, BlueQubit, and Algorithmiq -- to show how quantum calculations can be checked for accuracy in real time.
Historically, when a quantum computer tackled a problem too complicated for a classical supercomputer -- demonstrating what is known as quantum advantage -- scientists couldn't directly verify whether the quantum answer was correct. They had to extrapolate from smaller, simpler test runs, and assume the logic held at scale.
That meant researchers were taking answers on faith -- a risky bet for a technology still striving to prove its revolutionary potential to the public. In order to serve as reliable tools for scientific discovery, quantum computers must produce trustworthy outputs.
"I want to be very careful, because we're not claiming quantum advantage, " Jay Gambetta, International Business Machines' head of research, said in an interview with Barron's. "The more nuanced word is 'trusted computation.'"
His caution is warranted -- quantum advantage claims are notoriously difficult to benchmark. D-Wave Quantum claimed advantage last year using its signature annealing processor, only to face pushback from researchers who questioned how quantum computational speed is defined and measured against classical algorithms.
Alphabet-owned Google became the first to claim quantum advantage in 2019 using a random-number calculation. IBM famously pushed back, arguing that "supremacy" on contrived mathematical problems was misleading. Gambetta himself was among the researchers to challenge Google's claims in a blog post titled "On quantum supremacy."
IBM's measured language reflects the high stakes, but its achievement is undeniable. The company has successfully checked off one of the boxes on its technical roadmap: enabling the first examples of quantum advantage by 2026. IBM has effectively laid the groundwork, and now it's up to the broader scientific community to help develop a standard.
"We're in this new era where quantum advantage experiments are happening, " Gambetta explained. "These methodologies show you can trust the quantum computer. Then you can move to the next step, which is looking at useful applications where you can trust the output."
The quantum industry is striving to move past the science-experiment phase and achieve true commercial viability. Already, adoption of the technology is accelerating, as illustrated by a recent federal funding and policy push.
IBM itself operates the world's largest fleet of quantum computers over the cloud, selling access to clients largely in academia and the government. However, the business isn't a meaningful revenue driver just yet.
It remains unclear how much the quantum division will boost IBM's organic growth, a metric it has been trying to boost through software acquisitions. For now, quantum remains a jewel in IBM's crown, a moonshot project where its technical leadership is undisputed even as questions shadow other parts of its business.
Gambetta, who attended a White House summit earlier this month, has seen a shift in the way the technology is discussed. "Quantum is now becoming a scientific tool, and everyone is asking, 'How can I use it? How can I accelerate the application of it?'" he said. "Obviously, I want to see it go from a scientific tool to a business tool. But if we get the science first, then we can get the business."
The next major milestone on IBM's roadmap is Starling, a fault-tolerant quantum computer slated to launch by 2029. Looking ahead to the start of the next decade, IBM plans to scale to an even more powerful and precise machine called Blue Jay.