Quantum Builders
From Supercomputers to Quantum Accelerators: Building the Future of Heterogeneous Computing
A fireside chat with:
Laura Schulz
This article is based on a conversation between Laura Schulz and Daniel Rodán Legrain as part of the Quantum Builders series, sponsored by Qblox. Watch the full webinar for more on quantum HPC integration, the layers involved, and where the US and European approaches differ.
Laura Schulz is project lead for innovation at the Argonne Leadership Computing Facility, where she works on future heterogeneous systems and workflows. Before Argonne, she led quantum computing and technologies at the Leibniz Supercomputing Centre in Germany as part of the Munich Quantum Valley, and before that, she spent years at Lawrence Livermore National Laboratory working at the intersection of high-performance computing and industry-facing innovation.
Schulz traces her path into quantum HPC integration back to how HPC centers already operate. Facilities exist to track what's coming and have it ready before users ask for it, and quantum was no exception. "Somewhat naively, I kind of came from the perspective where I just didn't see it as a standalone," she said. "I just saw the whole advanced computational ecosystem, and okay, here's a great new thing that we want to look at."
Schulz breaks quantum HPC integration into distinct layers, starting with the physical reality of moving a device out of a physics lab. "You actually have the physical hardware," she said, describing systems that were "lovingly cared for by PhD grad students and technicians" now needing to survive inside a large HPC facility with its own shielding, power, and environmental demands.
Above that sits networking and connectivity, then system software. Quantum hardware doesn't behave like a CPU or GPU that HPC schedulers already know how to share. "It's a disaggregated accelerator, so you've got to deal with co-scheduling," Schulz said, describing the coordination needed so classical nodes don't sit idle while a job waits on the quantum side. A newer layer on top of all this is coordinating systems distributed across regional, national, or international networks. Her summary of the whole stack is direct. "Integration is not just plugging in a system and plugging an Ethernet cable to it and waiting for it to go," she said. "There's a lot involved here."
Asked whether quantum HPC integration is still proof of concept or closer to production, Schulz places it firmly in the first category, just a more informed version of it. "We are in proof of concept, but a lot smarter than we were when we were doing proof of concept a year or two or three ago," she said. She points to the shift in tone at recent conferences as evidence. "I was just at ISC a couple of weeks ago, and I was honestly a bit stunned at how much it was quantum this time," she said, contrasting it with early workshops where the field was still figuring out what questions to ask.
When asked about the biggest surprise in making this integration real, Schulz didn't point to a technical wall. She pointed out how much coordination the work demands across communities with different starting cultures. "There's a statistic from your PMBOK books that ninety percent of any project is communication," she said. Quantum hardware companies, software teams, and HPC operators are each evolving on their own timeline, and keeping them aligned takes more effort than expected. "Translation, communication, coordination, way more effort than even I think a lot of us thought," she said.
Schulz rejects the framing that either quantum systems need to become HPC-like or HPC centers need to become quantum-aware. Both have to give ground. Supercomputers are tightly tuned systems that can't simply absorb a new accelerator on its own terms. "You can't have a unique solution that works with your quantum system and then toss it over the fence to the HPC part," she said. "That doesn't work." At the same time, quantum hardware carries physics that HPC has never had to accommodate before. "There's gonna be learning and accommodation on both sides for all of this to work," she said.
At a recent workforce development session at ISC, Schulz put the question directly to a room split between physicists and HPC practitioners. The physicists wanted a better grasp of what HPC could do for their problems, not the internals of the compute itself, but where in their workflow it made sense to hand something off. The HPC side wanted the reverse. Schulz sees this as the real translation gap the field needs to close, one built around behavior rather than jargon on either side.
Comparing the two ecosystems she's worked in directly, Schulz describes a different entry point on each side of the Atlantic. Europe's approach, driven by EuroHPC, put early quantum hardware directly into HPC centers with a clear mandate. "Here's a quantum system," she said, describing that instruction. "The technology is still rapidly evolving, but go ahead and start figuring out how this is gonna become part of the computing ecosystem." The US side, shaped by DOE's mission-driven science culture, started further upstream, testing what quantum could do on specific applications before hardware was sitting in a facility. She sees the two as complementary rather than competing.
Schulz expects the shift to fault-tolerant quantum computing to unsettle some of what the field has learned during the NISQ era. "NISQ era was both good and bad," she said. "There's a lot that we learned, but there's a lot that we're probably gonna have to unlearn." How much of the current system software and scheduling work carries forward into fault tolerance is still an open question she's actively watching.
Asked who HPC centers find easiest to work with across the quantum value chain, Schulz points to companies that treat the problem as a whole system rather than a single component. She compares them to the integrators that already assemble HPC systems out of networking, storage, and compute vendors. "Those who are actually working on the hybrid solution are the ones that we want to encourage," she said, over companies that hand off a quantum system as a standalone box and leave the surrounding integration work to the facility.
Schulz's own read on the next decade comes back to something simpler than a roadmap. "The more infrastructure or ecosystem that we set up, the more mature it comes, the more interesting questions and results that are going to happen," she said. "I just want to see really great science."
That case for integrators over standalone hardware sits close to what Qblox builds toward, control systems designed to work inside the layered environment Schulz describes rather than as a box dropped into a facility. If your team is working through what quantum HPC integration looks like in practice, we'd be glad to talk.
Contact us to learn how Qblox supports quantum HPC integration.
