Building the future of computing, together

Ever since the Watson Scientific Computing Laboratory opened on the campus of Columbia University in 1945, IBM has thrived in a symbiotic relationship with its university partners. That lab eventually evolved into what we know of today as IBM Research. And over that time, IBM has found that technology researchers, students, and faculty all benefit when we define and approach the next era of computing challenges together.

Now, IBM Research is strengthening its alliances with some of the world’s top science universities to advance the future of computing. Through three major, recently renewed partnerships with ETH Zurich, MIT, and the University of Illinois, we’re accelerating the development of what’s next in quantum and classical algorithms and applications. This move reflects IBM Research’s deliberate strategy to define the future of computing, with each partnership contributing to that approach, while aligning with a shared vision: defining the hardest problems in computing and tackling them together.

The spirit of collaboration is woven into the DNA of IBM Research, and we cannot attempt to advance all technologies alone. We tailor our research agenda to drive ahead of the needs of the moment — something we can only do by working closely with top academics. And by focusing resources on universities that are aligned on the goal of advancing the future of computing, and which are already key drivers of impact through their wider network of science and engineering research institutions, IBM Research can amplify their effectiveness.

In a newly expanded agreement with Massachusetts Institute of Technology (MIT), we plan to advance the mathematics of computing to benefit areas like materials science, chemistry, and biology. With Zurich’s Eidgenössische Technische Hochschule (ETH), IBM continues a decades-long partnership with funding for professorships and new research into the algorithms that will make it possible for quantum and classical computers to work in harmony to simulate the natural world. And with the University of Illinois Urbana-Champaign (U. of I.), the IBM-Illinois Discovery Accelerator Institute will push forward research on quantum-centric supercomputing (QCSC) architecture, next-generation AI systems, and novel algorithms.

IBM Research is tackling some of the most challenging and high-risk problems in computing. By working closely with the world’s top engineering schools we will define and solve the next generation of scientific and technological challenges.

“Our goal is to create what’s next in computing, and this is done not in a vacuum, but together with the community, where we participate, we lead, and we collaborate with the leading researchers,” said Ruchir Puri, chief scientist at IBM Research. “Together we want to push the boundaries of what’s ahead in computing.”

The Cambridge connection

For nearly a decade, IBM has partnered with MIT to shape new frontiers of AI. In that time, the field has seen massive changes, most notably the Cambrian explosion of large language models and generative AI. Under the roof of the MIT-IBM Watson AI Lab, researchers from both institutions have been building large-scale AI systems that are more capable and efficient than what has come before. And this week, IBM Research signed a new 10-year agreement with MIT to continue this mission.

While the lab’s purview had previously focused exclusively on AI, we are now opening the aperture to include quantum and algorithms work. The planned research portfolio will focus roughly half on AI and half on quantum-related research under the newly inked agreement — with substantial overlap between the two. The AI component will include work developing new foundation models and researching how AI tools can be used for quantum computation. “Error correction will be the other one, as well as using quantum computing or quantum-centric supercomputing for scientific discovery,” said Hanhee Paik, head of academic research programs and director of IBM’s quantum algorithm centers. MIT’s program reflects the importance IBM Research places on the intersection of AI and quantum computing, added Paik.

It’s been an auspicious start for the lab at MIT — over 139,000 academic journal citations, nearly 500 joint MIT-IBM refereed publications, and an H-index is a measure of the productivity and impact of an individual's or journal's scholarly publications, or in this case an institution's. The number comes from the most-cited papers and their number of citations. An H-index of 170, for example, means 170 papers that have each been cited at least 170 times.H-index of 170 despite being a relatively young lab. But the numbers only tell part of the story.

David Cox, the IBM director of the MIT-IBM Watson AI Lab, emphasized that university partnerships like this one go far beyond just funding — they create an inherently collaborative environment. “We don’t just give money to universities and hope good things happen,” he said. Because IBM has such a strong presence on the MIT campus, students can get informal mentorship from IBM Research staff and benefit from their experience. “It’s a very high-bandwidth collaboration,” said Cox. “It’s more like an expansion of their community, than an interaction with a sponsor.” IBM researchers routinely serve on students’ thesis committees, he added.

This relationship is only possible because of the years of trust built through the MIT-IBM partnership since 2018. With this version 2.0 of the partnership, Cox is excited to dig in to the algorithms work and pour some gas on the fire of quantum computing. “We’re going to make it easier for material scientists, biologists, and chemists to leverage the exciting advances in quantum computing to make new scientific discoveries in their fields,” he said. “Putting AI, quantum, and algorithms work under one umbrella also creates exciting opportunities across these areas.”

Midwest expansion

About 1,000 miles west along I-90, another partnership is also expanding. Under a newly signed five-year research agreement, IBM Research and University of Illinois will expand the IBM-Illinois Discovery Accelerator Institute to usher in a new approach to high-performance computing: quantum-centric supercomputing. QCSC merges classical and quantum methods to accomplish what neither can do on its own, with CPUs, GPUs, and QPUs harmonizing via new workflow management tools and hybrid classical-quantum algorithms.

Since its 2021 founding, the Discovery Accelerator Institute has grown to encompass 20 active projects across hybrid cloud, AI, quantum computing, materials discovery, and sustainability, producing more than 230 research publications. This new phase of the partnership likewise spreads its focus to multiple areas but emphasizes system-level work. About half of the effort centers on AI-native systems research, particularly on novel AI architectures for fast-evolving data and AI workloads. This initiative also includes co-developing algorithms and AI systems for large-scale distributed inference. It’s what’s called an Algorithms-to-Silicon-to-Systems (or AS2) approach where algorithms, specialized silicon, and systems software are developed together to tackle classically hard problems in areas like chemistry, materials science, and next-generation AI workloads.

The other half is explicitly dedicated to QCSC.

In collaboration with the US National Center for Supercomputing Applications (NCSA), IBM and U. of I. are working together to provide QCSC resources to researchers on campus. The architecture knowledge developed under this collaboration is intended to be shared openly, enabling NCSA researchers to build, operate, and use the system for algorithms and applications research while contributing to the broader quantum computing community.

The expanded partnership between IBM and U. of I. aims to integrate NCSA’s Delta and DeltaAI supercomputers with IBM cloud-based quantum processors to enable QCSC demonstrations. “We’re trying to advance the next generation of computing systems,” Paik said.

This renewed deal also aims to put IBMers in closer contact with faculty and staff on campus, in much the same way that MIT students benefit from the immersion found there, added Sudhir Gowda, director of academic research collaborations at IBM Research.

A European union

Across the Atlantic in Europe’s most vibrant IT scene, IBM Research is deepening its relationship with ETH Zurich, one that has already spanned seven decades.

IBM and ETH's longstanding partnership has yielded groundbreaking scientific developments, from the Nobel Prize-winning invention of the scanning tunneling microscope (STM) in 1981, to more recent work this year building the world's first molecule with a half-Möbius topology and accurately simulating it with quantum hardware.

Under the new 10-year agreement, this partnership will be more tightly focused on algorithms and applications. The ETH program targets mathematically challenging and often intractable problems, such as eigenvalue problems, stochastic processes, partial differential equations, and optimization.

Researchers from IBM and ETH Zurich will blend quantum, classical, and AI-based approaches to develop new quantum and classical algorithms — rather than treating quantum computing in isolation. This holistic approach makes ETH a key partner for advancing rigorous, application-driven algorithm research. “Our collaboration with ETH is essential because of its world-class expertise in rigorous, application-driven algorithm research and its ability to translate bold ideas into transformative computational advances,” said IBM Research Zurich director Alessandro Curioni. With these algorithms, the team hopes to be able to run more complex simulations and effectively multiply the power of AI and quantum computing.

“IBM Research is one of the world’s best research institutions, but we cannot do everything ourselves,” said Paik. “That’s why university collaborations are so important to define and then explore the most challenging problems together.”