It’s an unscientific fact that nine out of every 10 conversations about tech in the past year have been about AI. But the 10th has been about quantum computing. In a period of just over a year, the perennial technology of the future has become suddenly real—with major breakthroughs in computing hardware and software, significant public and private investments, and rising stock prices for companies in the burgeoning quantum ecosystem.
In simple terms, quantum computers differ from regular, “classical” computers in the way that they encode information. While classical computers operate on a binary code of 1s and 0s, stored as bits, quantum computers use qubits, which can exist in multiple states of one-ness and zero-ness. Because the computer architecture is so different, a quantum computer solves problems differently—making mysterious leaps and bounds where classical computers proceed step by step.
It’s a common misunderstanding that quantum is all about making computers that are much faster—though it may do that in some cases. The real promise of quantum computing is that, by side-stepping the physical rules that define classical computers, it can go beyond their limitations—performing now-impossible simulations of atomic-level interactions in chemistry, medicine, and materials science, and enabling game-changing prediction of quantum-like phenomena such as weather and financial markets. Sufficiently advanced quantum computers—which could be developed by adversarial nation states like China within the next few years, according to many experts—would also obliterate current data-encryption algorithms. (In August, the U.S. Commerce Department’s National Institute for Standards and Technology released new “post-quantum” algorithms designed to counteract the threat).
The big news in quantum this month has been Google Quantum AI’s new quantum chip, called Willow, which set a new record in a benchmark performance test and demonstrated a promising approach to quantum error-correction that could be a breakthrough for scaling quantum systems. Willow took less than five minutes to complete a random circuit sampling (RCS) problem that would take one of today’s fastest supercomputers about 10 septillion years—longer than the universe has existed—to solve. The problem has no real-world application, but Hartmut Neven, founder and lead of Google Quantum AI, claimed on a company blog that the outcome of the test “lends credence” to the idea that we live in a multiverse.
Google is just one of the Big Tech players vying to lead the quantum computing race. It faces significant competition from the likes of Microsoft, Intel, and, not least, IBM. Last December, IBM debuted some of the most advanced quantum computing hardware the world had yet seen, including the Condor chip—a 1,121-qubit quantum processor that is the largest yet built—and the innovative Quantum System Two, a modular quantum computer that links together several of its smaller and less error-prone Heron chips. And the company is racing ahead on its quantum roadmap, which calls for delivery of a practically useful, fully error-corrected quantum computer in 2029.
But hardware is just half of what it takes to realize the goal of quantum advantage—that is, not just solving computing problems faster, but solving problems that can’t currently be computed at all. Quantum computers work in an entirely different way than classical computers, and they need a fundamentally different kind of software in order to do anything useful. The development of such software—the algorithms—is in its early days.
Over the past year, IBM has also doubled down on efforts to grow an open-source ecosystem that supports the development of new quantum software. Some 250 institutions and 600,000 users are signed up to remotely access the IBM Quantum Network. After major hardware upgrades, its U.S. Quantum Data Center in Poughkeepsie, New York now has more utility-scale quantum computers than any other location in the world. In October, the company opened its first European Quantum Data Center, in Germany. In March, IBM released major updates to its open-source quantum computing software toolkit, Qiskit, and in November held its first Quantum Developer Conference.
IBM and Illinois Governor J.B. Pritzker also recently announced a new collaboration to establish the National Quantum Algorithm Center, as part of the multibillion-dollar Illinois Quantum and Microelectronics Park that is expected to begin construction on Chicago’s South Side in 2025. The center will be anchored by an IBM Quantum System Two, and will enable researchers, national labs, and industry in Illinois to develop quantum computing algorithms for real, relevant use cases.
Jay Gambetta, IBM’s VP in charge of IBM’s overall quantum initiative, spoke with Fast Company about recent quantum headlines, the company’s long-term quantum vision, and the importance of being a team player in the evolving quantum ecosystem.
What did you think of Google’s announcements about the performance of their quantum computer, Willow?
Their device is impressive. They’ve made progress. Some of the metrics are better than ours; many are not. We’ve always stayed away from [making] artificial comparisons to large-scale calculations, like Google’s random circuit sampling experiment, which they have said has no practical value. And as far as we know, it’s just as hard to check that these quantum computations are correct using a classical computer as it is to perform the simulation. For this reason, comparisons between random circuit sampling demonstrations and the large-scale classical computations required to simulate them are artificial.
We’ve crossed the point that I call quantum utility, where we can run quantum computations that are too complex to simulate on a classical computer. This is an important milestone, but I think we want to get to “quantum advantage,” where you can do something you can’t do with a classical computer.
Google also made significant progress on “below threshold” error correction, showing that, as the number of qubits they used in Willow increased, the error rate decreased exponentially—the opposite of what usually happens.
I will acknowledge that they made progress in the device for error correction. What they’ve done is [based on] surface code [which uses adjacent qubits to “check” each other]. It’s a good demonstration. We know that the surface code, which IBM helped invent, cannot truly scale because of roadblocks including the significant physical qubit overhead and a monolithic design.
We’re focused on different types of codes called high-rate, low-density parity-check [which essentially “spot checks” widely dispersed qubits]. We’ve demonstrated a viable path to error correction using 90% fewer physical qubits, but require more “packaging” and a modular design. We’ve been working very hard on that—with new couplers that seam together adjacent chips. If we can put all these pieces together, you can build a code that, in my opinion, can actually be built. And we’re very confident we’ll deliver error correction by 2029.
So, you are still on track with the quantum roadmap you laid out in 2023?
We’re continuing to tick off everything we said we’d do. There’s hardware demonstrations of modularity, with Flamingo and Crossbill [462-qubit and 408-qubit systems that integrate next-generation connectors], which are significant steps towards our end goal of getting a fault-tolerant quantum computer working.
But what’s kind of exciting is that hardware and software are coming together—and the biggest updates that we have at this time are for Heron. By incorporating error-handling methods and things like that, we now have it running with accurate circuits of up to 5,000 two-qubit gates. This is something we said two years that we’d achieve and it is exciting because now, rather than looking at the noise in the system and doing things they already know the answer to, people are running things [on IBM’s quantum systems] where they don’t already know the answer.
In the past year or so, you’ve installed a quantum computer at the Cleveland Clinic and built the first quantum computer on a university campus at Rensselaer Polytechnic Institute in upstate New York. How are developers who don’t have a quantum computer on site actually interacting with IBM systems today?
Over 600,000 people have registered to use the IBM Quantum Network. And there has been a change in how they are using the system. The average usage last year was 13 qubits. The average usage now, with the Heron chip, is 112 qubits. This is huge. The whole community that’s using these systems has moved from [calculations that use] 13 qubits, which I could probably simulate with my mobile phone, to [ones that use] over 100 qubits, which I cannot simulate with the best supercomputers in the world. That transition happened this year.
Qiskit, our open-source SDK [software development kit] works with about 10 different quantum systems, including ours. It has been upgraded and benchmarked to be faster, easier, more reliable. Most developers, we don’t imagine, will actually want to get down to the low-level API, the quantum instruction set, and that’s why we created these abstraction layers to really simplify it so that more people can get value from it. We created this thing we call the Qiskit Functions catalog, which offers software from startup partners. As more and more people who are not from quantum physics want to apply quantum computing to their domain, there’s a shift from research for quantum to research with quantum. Over 3,000 papers have been published using our system.
What kinds of research are people doing?
The areas that I personally find exciting are things in calculating chemical structure. Cleveland Clinic has hired a very famous chemist, Kenneth Merz. He has pushed the limits of simulating molecular interactions in classical computers, developing algorithms that are important for future applications in drug discovery and things like that. He’s now taking his domain knowledge and applying it in the quantum space. Lockheed is exploring quantum applications in propulsion. The Riken Center for Quantum Computing in Japan recently published a paper using quantum computing to simulate interactions of iron and sulfur-type metals.
When will we get to quantum advantage, when quantum computers can do things that classical computers simply can’t?
To get to quantum advantage, where you can do something you can’t do with a classical computer, you need two things. First, you need the platform, and I think we’ve shown with the upgrades that the platform is completely there. The second thing you need is a community of domain experts doing algorithm discovery. Partnerships, in my honest opinion, are our differentiator. In the IBM Quantum Network, we have over 250 members who are scientists who are doing algorithm discovery. Now, with this National Quantum Algorithm Center in Chicago, we’re bringing together academics and industry together to really allow the scientific method to happen. To me, the scientific method is, you hypothesize a prediction, you test it, you verify it, and you iterate—and you can have this happen along with progress in the hardware. I’m optimistic that if we apply the scientific method to quantum advantage, we will see continual steady progress to quantum taking over what can be done with classical computing. I am very optimistic we’re going to see quantum advantage.
Autentifică-te pentru a adăuga comentarii
Alte posturi din acest grup
Back in 1979, Sony cofounder Masaru Ibuka was looking for a way to listen to classical music on long-haul flights. In response, his company’s engineers dreamed up the Walkman, ordering 30,000 unit
Even as the latest phones and wearables tout speech recognition with unprecedented accuracy and spatial computing products flirt with replacing tablets and laptops, physical keyboards remain belov
One of the most pleasant surprises about this year’s best new apps have nothing to do with AI.
While AI tools are a frothy area for big tech companies and venture capitalists, ther
The world of enterprise tech is built on sturdy foundations. For decades, systems of record—the databases, customer relationship management (CRM), and enterprise resource planning (ERP) platforms
Bluesky has seen its user base soar since the U.S. presidential election,
Russell Hedrick, a North Carolina farmer, flies drones to spray fertilizers on his corn, soybean and wheat fields at a fraction of what it
The emergence of generative artificial intelligence tools that allow people to efficiently produce novel and detailed online reviews wi
