Quantum Builders
The Quantum-HPC Frontier: Driving the Future of Hybrid Computing
A fireside chat with:
Bert de Jong
This article is based on a conversation between Bert de Jong and Daniel Rodan Legrain as part of the Quantum Builders series, sponsored by Qblox. Watch the full webinar for more insights on hybrid quantum-classical architectures, software stacks, and the workforce shaping the next decade of computing.
What happens when one of the world's leading computational scientists, someone who helped push classical supercomputing to the exascale era, turns his attention to quantum? And what does it really take to make quantum computers useful for science, not someday, but within the next five years?
In a recent Qblox Quantum Builders conversation, Bert de Jong, Department Head for Computational Sciences at Lawrence Berkeley National Lab and Director of the Quantum Systems Accelerator (QSA), laid out a pragmatic, deeply informed vision for how quantum and high-performance computing will come together. His perspective is shaped by decades of building tools that solve real scientific problems.
The takeaway: Quantum won't replace HPC. It needs HPC. And getting the integration right, across hardware, software, and people, is the defining challenge of the next five years.
De Jong’s career began in high-performance computing and computational chemistry, driven by a simple motivation: the problems he wanted to solve demanded enormous computational power.
"I started in HPC, worked a lot on exascale computing, simply because I want to solve scientific problems and the problems I want to solve require a lot of computational power," he explains.
When he arrived at Berkeley Lab about 12 years ago, he started looking beyond exascale. The question wasn't just how to make classical computers faster. It was what comes next. That led him to quantum computing, not as a replacement for HPC, but as an additional tool in the scientific toolkit.
"I see HPC, quantum, and AI as tools to solve real scientific problems that are relevant to the Department of Energy."
His teams began working on quantum early, focusing on a practical question: how can we actually use quantum computers to do useful scientific simulations, starting with chemistry and expanding into materials science, high-energy physics, and nuclear physics?
One of de Jong's most striking observations, and one that resonated deeply with the audience, is about the current relationship between quantum and classical computing.
"I don't think quantum is going to fully replace HPC, if anyone has ever been worried about that," he says. "If you look at it right now, quantum actually needs HPC. And more than HPC needs quantum, to be blunt."
The reasons are concrete. Compiling quantum programs is computationally expensive. It's an NP-hard problem, and current compilers rely on heuristics that demand significant classical resources. Error correction, whether done in real time or after the fact, requires substantial processing power. And the data produced by a quantum computer represents a vast amount of information that requires classical post-processing.
"Pre-processing, post-processing, potentially error correction: they will always need some level of HPC," de Jong notes.
But the integration won't look like a GPU bolted onto a CPU. The physics is too different, the data flows too loosely coupled. De Jong envisions quantum computers as part of an HPC center's offering, but integrated more loosely than accelerators, handling specific chunks of computation that are exponentially hard for classical machines and then handing back the results.
De Jong is refreshingly direct about what quantum computers are and aren't good at.
"A quantum computer is not a good calculator. It doesn't know how to do one plus one well. It will not do your word processing and email."
So, where does quantum have an edge? The answer traces back to Feynman's original insight: quantum problems need quantum machines.
"We can never translate a quantum problem to a classical machine in the most efficient way. We know that in chemistry. If we want to solve a problem exactly, we need exponentially large resources. This is where a quantum computer can play a role."
The early wins, in de Jong's view, will come in quantum physics, chemistry, materials science, drug discovery, and certain optimization problems. Computational fluid dynamics and finance are showing early potential as well. Finance in particular stands out, because if quantum can deliver even a marginal speed advantage, the economics of high-volume trading make adoption almost automatic.
He's more skeptical about AI on quantum computers in the near term. The fundamental bottleneck is data movement: quantum computers excel when a small amount of data goes in, an exponentially hard computation happens, and a small amount of information comes out. AI workloads, by contrast, require massive data ingestion, and every measurement on a quantum computer collapses the state, forcing a restart.
"Getting large amounts of information in or getting large amounts of information out is an exponentially hard problem," he explains. Solving that will require fast, rapidly loadable quantum memories, a technology that doesn't yet exist at the needed scale.
If you ask de Jong about the biggest practical barrier to useful quantum computing, his answer keeps coming back to software.
"I always joke that not too long ago we were just assembly programming a quantum computer, which is how we did the early classical computers," he says. "Remember, classical computers have been around for 50 or 60 years. They have had the time to develop the software stacks, understand how to make easy-to-use software, and make things seamless for an end user. Quantum is just barely starting to get there."
The challenge runs through every layer. Compilers today translate scientific problems into circuits and then into assembly language, but they're often not tightly coupled to control hardware below or to applications above. There's no equivalent of Python's just-in-time compilation ecosystem. And the diversity of quantum hardware (superconducting qubits, trapped ions, neutral atoms, photonics) makes standardization difficult.
De Jong sees the compiler as the critical middle layer, but argues it needs to reach in both directions: down to the control hardware, and up to understand the application.
"Whatever happens at the control level, if we know what's happening at the application level, we can make better choices at the compiler level to build the most optimal program," he says. "Just like we're getting physics-aware AI, we kind of need physics-aware compilers."
De Jong currently leads Mach-Q (short for "speed of quantum"), a DOE software center that evolved from an earlier program called AQC (Aiding Quantum Computing). Rather than building another end-to-end stack, Mach-Q focuses on identifying and filling gaps in the existing software landscape.
One of its flagship products is BQSKit, a compiler infrastructure built at Berkeley that has become one of the leading quantum compilers. The team is now re-engineering it to be error-aware, capable of producing error-corrected quantum circuits by incorporating detailed information about where and how errors occur on specific hardware.
Beyond the compiler, Mach-Q is also building tools for hybrid quantum-HPC integration and exploring a provocative idea: multi-modality quantum systems. Rather than choosing one qubit technology, future systems might run different parts of a problem on different types of quantum hardware simultaneously.
"You could potentially think of quantum computing systems that might have multiple modalities together, where certain parts of the problem are run on one type of quantum computer and some parts are run on another," de Jong explains.
The team's philosophy is plug-and-play. Rather than offering a monolithic stack, they're building modular components that slot into existing workflows and fill specific needs.
When an audience member asked whether a standardized assembly language is a prerequisite for higher-level quantum software, de Jong offered a nuanced answer.
"A uniform assembly language would be a little hard right now," he says. Unlike classical computing, which eventually converged around x86, quantum hardware modalities are still fundamentally different from one another. Atoms and ions behave differently from photons, which behave differently from superconducting qubits.
The more promising path, in his view, is standardization one layer above assembly: quantum intermediate representations (QIR) that allow compilers to target multiple hardware platforms without needing to know every detail of the underlying control system.
"It would be good to get to a point where we can get some unification at the QIR level or some level above assembly, that really brings the community together and allows us to build APIs at different levels going forward."
A recurring theme in the conversation is the unique role of national laboratories in the quantum ecosystem. De Jong sees a clear division of labor.
Academia excels at fundamental science with long time horizons. National labs serve as a translational layer, taking those fundamental advances, engineering them to scale, and raising them to higher technology readiness levels. Industry then takes those prototypes and commercializes them, building systems at scale in a cost-effective way.
"It's the commercial entities that are going to take the components, build at scale, and then make it a commercial product that can be made available to those who would utilize quantum computers in the future."
A concrete example: through the Quantum Systems Accelerator, Sandia National Laboratories has been fabricating integrated photonic circuits, a technology critical for scaling atom-based quantum systems. After five years of high-risk research that would have been too expensive for most startups, the technology has reached a point where the industry can adopt and integrate it.
"Doing that kind of research is expensive, time-consuming, and high risk," de Jong says. "But most of the atom-based technologies know they need that technology in the future. They're living off investor dollars and have a road map of two to three years to deliver something. Integrated photonics might be a little further out. This is where the national labs can take that high risk."
Having grown up and earned his PhD in the Netherlands before moving to the US, de Jong brings an international perspective to the conversation. And he's candid about where the US falls short.
"Europe seems to have a good grasp on how they can maximize the commercialization of this technology in the future," he observes. "The US has been honestly a little bit more fragmented."
He points to several European examples. Denmark, with backing from Novo Nordisk and the Danish government, has procured a neutral atom system through a public-private model. The UK's National Quantum Computing Centre houses multiple prototype systems to give researchers broad access. The Netherlands and other countries have developed comprehensive strategies that link research goals directly to commercial pathways.
In the US, by contrast, government agencies like DOE and DoD pursue their own programs while a large ecosystem of commercial entities operates independently. The result is less coordination than de Jong sees across the Atlantic.
"I think this is a lesson for us to definitely learn: make sure that industry, national labs, government, and academia get together, work closely together towards that singular goal."
Like many leaders in the field, de Jong sees workforce development as one of the most urgent challenges. But his vision extends well beyond training more quantum physicists.
"Once we're starting to think about deploying systems, making them available, assembly... you start to think about a very different category of people. It's not the scientists and engineers anymore. It's actually technicians."
The quantum industry increasingly needs people who can do cabling, assemble cryogenic systems, tune lasers, and support large-scale deployments. These are roles that require associate degrees or bachelor's degrees, not PhDs.
"We can go to a community college right now and talk about quantum and they're like, 'I didn't know that was even a pathway for us.' It was too esoteric. It's quantum physicists. No, it's changing. It's now an engineering and technician role."
To address this, the Quantum Systems Accelerator launched QCAMP, a program that brings quantum concepts to high school students and, critically, their teachers. Starting in California and New Mexico, it has expanded to eight or nine states, with de Jong's long-term goal of reaching all 50.
"A teacher is the one that's going to inspire the students," he says. "We are trying to teach teachers, especially in smaller rural communities, what the concepts of quantum are so that they can teach it to some level in their high school classes."
The next phase will extend into community colleges and bridge programs designed to retrain technicians from other industries.
For classically trained computer engineers or technicians interested in entering the quantum field, de Jong's advice is encouraging.
"You don't have to be proficient in quantum physics or quantum mechanics, but you need to have some level of affinity to it," he says. At its most fundamental level, quantum mechanics is matrix-vector multiplication. Once classical engineers see the connection between what they already know and how quantum systems operate, the transition becomes natural.
He points to two books his teams use regularly. The first is "Quantum Computation and Quantum Information" by Nielsen and Chuang (affectionately known as "Mike and Ike" in the Bay Area), of which the first three or four chapters provide a solid foundation. The second is a more recent text by Tom Wong that bridges classical and quantum computing concepts.
But de Jong emphasizes that much of the day-to-day work in quantum doesn't require deep knowledge of quantum mechanics at all. Tuning lasers, cooling a fridge to 10 millikelvin, installing hardware: "That is a simple engineering problem, and there are a lot of lower degrees, AA degrees, bachelor of science degrees, that can be critical to get this done efficiently."
When asked what milestones will signal that the field is on the right track, de Jong focuses on three metrics that the industry must watch closely: time to solution, accuracy, and cost.
"Those are three metrics that the industry has to closely watch and articulate as the pathway to beat, for example, HPC or beat AI," he says. "The cost point of AI is pretty high right now. The cost point of HPC is pretty high. So I think there is a quantum advantage from speed and performance, potentially, but there are more stories to be told about the ability to do scientific simulations and solve key industrial problems in a different way that has a lower cost profile."
In the near term, the field needs demonstrations of value on current and next-generation systems. De Jong believes the hardware arriving in the next few years will be capable of showing meaningful results that matter to end users, government agencies, and industry partners.
"We really need to demonstrate value. And I think the systems that are there right now, the next level of systems I can see on the horizon, are going to be systems that can show value."
Longer term, the goal is producing quantum systems at scale, because volume is the only way to drive costs down enough to deliver on the promise of broad societal impact.
De Jong closed the conversation with a message that doubles as a call to action.
"That motto is definitely the key part that will get quantum to the point that it's going to be a fundamental building block of our society. It has to be together. It has to be a team. I understand industry is competing, but I think we're all in it together, and we're all going to be better for it once we deliver what the potential is of quantum computing going forward."
The next five years will be decisive. The hardware is reaching a tipping point. The software gaps are being identified and filled. The workforce pipeline is expanding. And the collaborations between national labs, academia, and industry are deepening.
The question is no longer whether quantum will matter. It's whether the ecosystem can come together fast enough to realize its potential.
Quantum progress depends on collaboration across disciplines, institutions, and the boundaries between research and industry. At Qblox, we're proud to support that vision. Whether you're exploring new qubit modalities in the lab or scaling toward production, we're here to help you build what's next. Contact us today for more information.
