Shaping the next era of agentic AI at Think 2026

4 p.m. EDT | May 6, 2025IBM executives used the company’s Think conference this week to argue that quantum computing is moving from experimental science to commercially relevant applications.The announcements came on the last day of the conference, as technology companies attempt to prove that quantum systems can solve commercially meaningful problems after years of scientific breakthroughs and billions in investment. On stage with its partners, IBM unveiled new results that executives said pushed quantum systems beyond research demonstrations and into practical work involving drug discovery, materials science and fusion energy.“Useful quantum computing is here right now,” Director of IBM Research and IBM FellowJay Gambettasaid on stage.He framed industry’s next phase around what IBM calls “quantum-centric supercomputing,” an architecture that combines CPUs, GPUs and quantum processing units (QPUs) into tightly integrated systems. Instead of replacing classical computers, the systems divide calculations between conventional and quantum hardware, depending on which machine handles a specific task best.

IBM pointed to a series of early scientific use cases to support its argument that quantum computers are beginning to tackle problems that strain conventional systems. One of the keynote demonstrations came from researchers at Cleveland Clinic, who described using IBM quantum systemsto modelincreasingly large biological molecules tied to pharmaceutical research.Kenneth Merz, who leads a lab at Cleveland Clinic that collaborates with IBM on hybrid quantum-classical computing models for biomedical research, said that his team began with relatively small molecular calculations before scaling up to protein systems with more than 12,000 atoms.“We did this, and we thought we were really going big, but we quickly realized we could have been two or three times larger,” Merz said.Researchers used a technique calledSQD, short for sample-based quantum diagonalization, to estimate molecular behavior at scales that classical methods increasingly struggle to handle efficiently.The company also highlighted arecent projectinvolving a newly synthesized “half Möbius” molecule, a structure shaped like a twisted loop that does not naturally occur in nature. IBM researchers used quantum systems to simulate the molecule’s electronic structure and compare it with experimental imaging data. Gambetta said the results showed quantum hardware helping interpret real laboratory measurements rather than isolated benchmark exercises.

IBM and Oak Ridge National Laboratory researchers also talked about a project to use quantum systems to study fusion reactor fuels, a key challenge in efforts to develop commercially viable nuclear fusion energy. The effort combined AI, high performance computing and quantum processors to model the chemistry involved in tritium production. Scientists said the chemistry remains difficult to model accurately using conventional computing methods alone.Sarp Oral, Section Head for the Advanced Technologies Section and a Distinguished Research Scientist at the National Center of Computational Sciences (NCCS) Division of Oak Ridge National Laboratory, told the session that the project combined AI, high-performance computing and quantum processors into a coordinated workflow aimed at accelerating scientific research.“What we are trying to build in this workflow is to bring together different computing paradigms,” he said.One of the session’s biggest announcements came from quantum computing startupQ-CTRL, which shared that it hadachievedwhat it called “practical quantum advantage” using IBM quantum hardware. The company’s CEO and Founder,Michael J. Biercuk, said that the companies solved a commercially relevant materials science problem using a 120-qubit IBM quantum system: the quantum workflow completed the calculation in roughly two minutes, while a comparable classical simulation took more than 100 hours on a computing cluster.“That’s a 3,000 times speedup owing to access to quantum computers,” Biercuk said.