AI, quantum and high-performance computing join forces at SC25

Supercomputing has shifted into the mainstream, with recent AI advancements and adoption making the use of massive amounts of computational power more commonplace. The result is a thriving ecosystem that blends historical supercomputing players with their cutting-edge technology counterparts in AI, photonics and quantum computing. At SC25, 2025's largest conference for super- and high-performance computing, held this year in St. Louis, the compute ecosystem of the future began to take shape, with shared capability and long-term potential the twin threads tying once-disparate disciplines together.

As individual supercomputing functions mature and quantum computing and AI continue to improve in functionality, the next step in enterprise adoption is to determine how all these rapidly progressing pieces will work together. While AI is tackling many complex problems, the addition of quantum technology could address new classes of computing tasks, particularly as quantum computing continues to scale. Quantum computing is not part of your typical computing portfolio, but techniques like quantum annealing are finding practical applications today, and pure-play quantum approaches are pushing ever closer toward quantum advantage — the point at which quantum computers are doing things that classical computers simply cannot. Progress isn't constrained to just quantum and AI: New areas like photonic computing and improvements to underlying infrastructure are all jockeying to be part of, and enable, the future of computation.

Supercomputing swings through St. Louis

The 2025 edition of the International Conference for High Performance Computing, Networking, Storage and Analysis, otherwise known as SC25, was held in St. Louis, with over 16,000 attendees and a record-breaking 559 exhibitors. The theme for the event was "HPC Ignites," with the conference focusing heavily on the emerging and expanding technologies poised to revolutionize the future of computing. In line with that focus, topics including agentic AI, quantum computing, photonic and optical computing, and more took to the floor (and the stage) alongside more traditional HPC vendors and interest areas. Model quantum computers were displayed across the aisle from plumbed cooling systems, and the world's largest technology vendors mingled with academic consortia pursuing the next big scientific advancement.

Quantum computing boasted a large presence at the event, with a section of the main exhibition floor dubbed the "Quantum Junction," and dedicated to the intersection of quantum and HPC. Quantum technology vendors including Alice & Bob SAS, Alpine Quantum Technologies GmbH (AQT), Bluefors Oy, Classiq Technologies Inc., D-Wave Quantum Inc., Infleqtion, IQM Quantum Computers, Keysight Technologies Inc., Pasqal SAS, Q-Ctrl Pty Ltd, Quandela SAS, Quantinuum Ltd., Quantum Computing Inc., Quantum Machines, QUDORA Technologies, QuEra Computing Inc., Qunova Computing Inc., Quobly and Zurich Instruments AG were among the exhibitors, along with various quantum-enabled research institutions and consortia, including Juelich Supercomputing Centre, Leibniz Supercomputing Centre, Open Compute Project Foundation and RIKEN.

Key themes and takeaways

At the weeklong conference, several key themes surrounding quantum, AI, and the future of compute emerged from the panels, discussions and conversations held throughout the venue.

Hybrid computation is emerging as the likely future of compute. One of the clearest signals sent by the increased presence of quantum computing vendors at SC25 is that quantum is positioning itself as part of computing's future — a future that is likely going to be hybrid, mixing classical compute, AI processing and quantum computing to solve new problems. Early hybrid workflows are beginning to take shape, with projects embedding quantum processors in HPC clusters, utilizing GPU-accelerated error mitigation and signal processing, and even creating quantum simulation frameworks that could run across CPU/GPU/QPU pipelines simultaneously. While the discussions at SC25 remained primarily focused on proof-of-concept work and problem-solving, the timeline for hybrid computation continues to move up.

Quantum computing has entered the world of supercomputing in a big way, positioning itself as a "third space" alongside classical compute methods. In a social context, third spaces are environments outside the two primary settings of daily life (home and work/school). If we think of classical computing (CPUs) as the first space and AI/GPU-centric accelerated computing as the second, quantum computing is increasingly filling the "third space" role as an environment that is unique, slightly quirky and more fluid than its counterparts. Additionally, just like social third spaces serve as a place for different groups to mingle, quantum computing is full of experts in AI, physics, HPC, mathematics, materials science, finance, chemistry and more, becoming a meeting ground where early experimentation and a blending of ideas is sparking novel conversations and solutions that might not emerge in traditional compute environments.

The compute ecosystem is becoming a story of cohesion, as partnerships and collaborations among disparate compute players emerge and evolve. NVIDIA Corp., for example, had a large presence at SC25 and has been especially active in quantum over the past few months, expanding its partner network, launching a quantum research center in Boston and most recently unveiling its NVQLink platform to stitch together quantum-classical workflows. Amazon Web Services Inc. also had a notable presence at SC25, particularly within the quantum computing space. At one point in the event, the AWS booth was "taken over" by some of the company's quantum computing partners, with the hyperscaler hosting dozens of quantum-centric presentations and displays from across the quantum ecosystem.

Increased collaboration in the quantum computing space is notable because some of the most influential players in quantum right now aren't pure-play quantum companies — instead, they are HPC and AI giants accelerating the ecosystem from the outside in. With quantum partnerships increasingly expanding outside of the historical quantum ecosystem, the quantum landscape is becoming more interconnected with other disciplines than ever.

Photonic or optical computing was an unexpected, but notable, topic at the event this year. In photonic computing, the hardware and architectures use light (photons) to carry, process and/or store information. In contrast, classical computing is almost exclusively electronic, relying on the movement of a charge through transistors. By making use of photons, optical computing offers the potential for fast, energy-efficient computation ideal for high-bandwidth and parallel computations. Photonic computing, while decades old in theory, has seen a resurgence in interest and activity over the past year, driven by the explosive growth of AI workloads and data center demands.

While fully optical digital logic (i.e., optical transistors) is still largely experimental, SC25 saw a slew of companies offering analog optical computing products along with photonic accelerators for linear algebra and AI/machine learning use cases. As with quantum computing's quantum processing units, there is the potential for optical computing to integrate with other processing units in a hybrid computational approach: a mix of CPUs, GPUs, QPUs and LPUs (light processing units) — all working collaboratively for maximum efficiency, particularly on HPC and supercomputing-scale projects.

The software story is the next frontier. While much of the focus at SC25, and in quantum computing more broadly, has remained on hardware, the narrative is beginning to shift toward software tools and usability. Over the next three to five years, as quantum processors become broadly useful within hybrid workflows, progress will hinge on the maturation of software layers that translate hardware's theoretical capability into practical deployment. Platforms that emphasize high-level quantum algorithm design and automated circuit synthesis, such as Classiq, exemplify this shift toward abstraction and accessibility, and point toward a future where quantum development becomes approachable for today's software engineers. In a nod to that goal, the Classiq team at SC25 used its platform to power a quantum random number-based raffle, a modern take on a conference classic that also turned an abstract capability into a tangible, user-facing experience without requiring direct engagement with quantum complexity.

Supporting infrastructure

Alongside the computational finery was a significant amount of the latest in support infrastructure, from cooling to racks to enclosures in which to house it all. As power densities increase, there are requirements for increasingly capable and exotic systems to provide power and remove the heat that all of that power generates. HPC environments were once uncommon and niche applications, but AI clusters are now being put in place by many enterprises. With even moderately sized clusters dissipating multiple megawatts of energy, higher-capacity cooling technologies are required. That means that chilled air is giving way to liquid cooling, often delivered directly to the GPUs and ASICs doing the work. SC25 showcased the latest in cold plate technologies for direct-to-device cooling and offered advances in two-phase cooling, where the coolant boils off on the plate for even greater heat-rejection capacity. There were also immersion and spray cooling products, enabling an entire circuit board or chassis to be bathed in coolant.

On the power side of the equation, the shift to high-voltage DC power distribution is well underway. Traditional data center designs moved AC power throughout a facility via step-down transformers. Hyperscalers have been using DC distribution to reduce power losses in transformers, and now that change is making it to more mainstream designs. It is more evidence that stepping up to this level of computational power comes with a set of supporting technologies that are new to the average enterprise and require specialized skills to operate effectively.

What to watch next

SC25 made one thing clear: The future of computing will not be defined by any single breakthrough technology, but by how effectively disparate systems are woven together. AI, quantum, photonic and classical computing are no longer evolving in isolation, but are instead converging into a single, cohesive ecosystem. Looking forward, the next questions to be answered lie in the development and maturation of the software, infrastructure and collaboration needed to support an integrated computational future.

This article was published by S&P Global Market Intelligence and not by S&P Global Ratings, which is a separately managed division of S&P Global.

S&P Global Market Intelligence 451 Research is a technology research group within S&P Global Market Intelligence. For more about the group, please refer to the 451 Research overview and contact page