IBM Research and ETH Zurich open a new era of innovation

IBM Research and ETH Zurich just signed a new research agreement, with the goal of creating the algorithmic underpinnings of the next era in computing. This partnership has been longstanding, with the IBM-ETH relationship dating back 70 years, to the very start of IBM’s presence in Zurich.

The Zurich lab’s inaugural director, Ambrose Speiser, was a young computer science professor at Eidgenössische Technische Hochschule (ETH), hired by Thomas Watson Jr. himself to build a premier European lab staffed by the brightest minds on the continent. “It turned out – not unexpectedly – that the new IBM laboratory in Zurich was an attractive proposal for young engineers looking for a position,” Speiser later wrote in the IEEE Annals of the History of Computing. “Accordingly, it was not difficult to assemble a group of capable people.”

Capable was an understatement. Far from IBM’s main research lab in New York, IBM Zurich grew into a scientific powerhouse in its own right, with two Nobel Prizes to its name and discoveries that helped pave the way for nanoscale semiconductors and electronics, as well as high-performance scientific computing.

IBM’s close tie to ETH Zurich has been part of the winning formula. ETH supplied fresh talent to IBM, and sometimes the reverse, as seasoned researchers left IBM to teach at ETH, ensuring a steady exchange of ideas between academia and industry and applied science.

The lab’s outsized reputation drew Alessandro Curioni, then a PhD student at Italy’s elite Scuola Normale Superiori in Pisa. He came as an intern to IBM Research Zurich, then a hotbed for computational chemistry, and never really left, except to return to Pisa to finish his PhD.

As a researcher, he applied powerful computers to scientific problems, eventually earning the prestigious title of IBM Fellow for his work. In 2016, Curioni moved into management as VP of Europe and Africa, and head of the Zurich lab where he had started as an intern two decades earlier. We recently sat down with him to talk about IBM and ETH’s next chapter, and what an ‘AI times quantum’ vision for the future of computing looks like.

The Zurich lab turns 70 this year — funny to think it all began with an ETH professor! It’s been a continuous story of exchange, interaction, and success on both sides, by institutions that are unique in the world. ETH Zurich is one of the best universities in Europe, with 22 Nobel Prizes, including one that went to Albert Einstein, its most famous alumnus. And IBM is one of the oldest, most enduring tech companies in the world, with six Nobel Prizes and 7 Turing awards. It’s important to look at this as only a new chapter.

What will this next chapter focus on?

We’re creating the algorithmic basis for the future of computing where quantum information theory and classical information theory come together, extending our current applications and enabling entirely new ones. This fundamental shift will impact 95% of the problems we’re trying to solve in science and in business. IBM’s investment will fund students and projects in classical and quantum algorithms.

You came to the lab in 2006 as an intern — what brought you there?

IBM Research Zurich had a collaboration with ETH under the structure of the C4 Community for Computational Chemistry. I came during my PhD because it was the leading place to do computational chemistry. When I got to Zurich, I found the best place in the world to do my research.

When you became a full-time scientist, what did you work on?

At the time, I had classical instruments to simulate molecules in an approximate way. I was using the first IBM Power machines in the world to do science with ETH and C4. Then we went to massively parallel computers trying to exploit classical computers to do quantum simulation, hitting all the possible walls. Fifteen years ago, I started with IBM Blue Gene and massively parallel computing to try to address computationally interesting problems outside of chemistry. We worked together to create a fluid dynamic application simulating clouds of collapsing bubbles under pressure and won the 2013 ACM Gordon Bell prize.

And then a second prize two years later!

The 2015 ACM Gordon Bell Prize was for another HPC project, this time simulating flows in Earth’s mantle to understand the dynamics that produce earthquakes.

What other projects has IBM Research led with ETH Zurich?

We built a nanotechnology center in 2011. It was the first big public-private investment in Switzerland, and it sprung from the scanning tunneling microscope (STM), which was invented in our lab by IBM physicists Gerd Binnig and Heinrich Rohrer, who was himself an ETH graduate. They both received a Nobel Prize in physics in 1986. Their work was groundbreaking because it gave us a way to visualize and manipulate nature, atom by atom, and led to the rise of nanotechnology.

What’s the best part of working at Zurich?

It’s off the charts for innovation and quality of life. IBM was the first tech company to arrive here, adjacent the ETH campus. Today there’s Google, Meta, OpenAI, and more. The city still attracts the best minds from around the world. A few years ago, ETH was able to get Alesso Figalli, the mathematician who won the Fields Medal in 2018. You work here because of the ecosystem.

What excites you most about the future?

Quantum will reduce the computational complexity of simulating nature enormously. It will take away the bottleneck that limited the work we could do in the past. We were using classical computers to simulate the same things we’re trying to simulate today, but the computational complexity was so bad that when you enlarged the system or increased the accuracy, you hit a wall and couldn’t go on. It was frustrating as a scientist. You wanted to change the world, but you could not.

When you repeatedly hit the wall on your projects what kept you going?

I focused on making incremental improvements. I did this for years, starting with simulating molecules, then molecules in solvent, then molecules in solvent on interfaces, to get something better and more accurate. But I felt like Don Quixote fighting the windmill. I would push the wall a centimeter at a time, waiting for the magic to happen.

What made you decide to become a manager?

Once I achieved IBM’s highest technical honor, IBM Fellow, I felt like I needed a new challenge. It had been my dream to win the Nobel Prize since I was 20 years old, after hearing on the news that researchers at IBM Zurich had won the prize for the STM. At some point I realized that my probability of winning a Nobel was not so high. I thought I could do more for the company and the world by pushing the younger generation along. When Arvind [Krishna, then Research Director] asked me to lead the Zurich lab, I took just one night to say yes.

How will AI shape the future of quantum research?

The future won’t be AI plus quantum, but AI times quantum because quantum computing will be broader than it is today. Quantum will be an enabling technology. Classical computing will be part of quantum computing just as classical mechanics is part of quantum mechanics. This quantum computing engine will make AI more powerful. AI, meanwhile, has already made quantum more powerful.

How so?

We’re using today’s AI to invent new quantum and classical algorithms and implement them in machines more effectively. AI can help us run complex simulations in an autonomous way. Once quantum becomes the base of computing, the AI running on top will be even more powerful. AI times quantum will have a multiplier effect.

Where will ‘AI x quantum’ have the biggest impact?

Quantum mechanics was initially conceived to explain physics problems that classical mechanics could not. How something could be a particle and a wave at the same time. Later, we realized that quantum mechanics superseded classical mechanics. We came to quantum computing by realizing that some problems were too computationally inefficient for classical computers. Simulating nature is one of them. We’ve been talking forever about digital twins, and building accurate models of nature, but the digital twins so far have been bad carbon copies.

If you could simulate some aspect of nature with a quantum computer, what would it be?

One of the problems where I hit the wall was simulating batteries. We worked on a project to create new lithium-air batteries to revolutionize energy storage. When quantum computing matures, I think we’ll be able to accurately study these processes. To simulate a device like a battery and create real digital twins, we need to bring Hamiltonian simulations and stochastic processes together.

So that’s where ‘AI x quantum’ comes in?

Exactly. With the best possible classical and quantum algorithms we can multiply their power. That’s why this collaboration is so exciting. How many places in the world can you say that you have the technology and the skills, but you also have the legacy? IBM and ETH planted the seeds for nanotechnology and computational chemistry and grew these fields together. Now, we can do it again for computing.