GSK's Anthony Rowe Puts Useful Quantum Chemistry at About 2,000 Logical Qubits Around 2030; Cleveland Clinic's Lara Jehi Tells Hospitals to Prepare Now
Quentir Medicine Monitor
Evidence-based insights for quantum medicine. Published by Quentir Systems LLC · October 19, 2026.
Anthony Rowe, who runs research data platforms, scientific computing and AI infrastructure at GSK, gave a drug company's answer to the oldest question in quantum computing: how big must the machine be before it helps? Speaking at the City Quantum & AI Summit earlier in October, he put the tentative threshold at about 2,000 logical qubits for chemistry problems he judges beyond conventional computers, with 2030 as the horizon on current hardware roadmaps.
On the same stage, Lara Jehi, chief research information officer at Cleveland Clinic, described how her hospital, IBM and the quantum software company Algorithmiq modelled a light-activated cancer drug with a hybrid quantum-classical workflow, work that won a $2 million prize from Wellcome Leap in April 2026. Her message to hospitals was to take part now, while the machines are still being designed.
Computer Weekly reported both talks on 19 October 2026. Read together, they give a hospital research office and a pharmaceutical buyer two useful reference points: a working demonstration at today's scale, and an industry estimate of the scale at which the chemistry starts to pay. The distance between the two is the planning problem.
What GSK's Anthony Rowe said about 2,000 logical qubits, 2030 and his own team's move to AI
Rowe's figure came from his earlier pilots in quantum chemistry, protein folding and optimisation before he joined GSK. According to Computer Weekly's account of the summit, he called the threshold tentative and acknowledged that better algorithms or hardware could change it. A logical qubit is an error-corrected unit built from many physical qubits, so 2,000 logical qubits implies a machine with a far larger number of physical ones.
The more candid part of his talk concerned people. Rowe had assembled a team to explore quantum applications, and its members asked to work on AI instead. "They all came to me and said, 'We'd rather work on problems in AI right now because we can be more successful.' So fine. I mean, we can pivot," he said, adding that the industry side has to keep generating a return on investment. He also argued that the computational chemists and physicists already working in life sciences have skills that would transfer to applied quantum computing as the hardware matures.
Rowe's account describes his own experience and his team's choice. It does not describe GSK's overall research budget, and the 2,000 figure is a speaker's estimate given in a conference talk. It should be read with that weight.
How the 2,000-logical-qubit figure compares with IBM's roadmap: Starling in 2029, Blue Jay in 2033 or later
Rowe did not say which roadmap he had in mind, so one public roadmap serves as a check. IBM's published quantum roadmap places Starling, its first fault-tolerant computer, in 2029 with 200 logical qubits running 100 million gates. The company described the Starling design in a June 2025 research post, with construction planned in Poughkeepsie, New York. The step that reaches Rowe's number is Blue Jay, which IBM says will run circuits of one billion gates on up to 2,000 qubits in 2033 or later.
On that one vendor's schedule, the stated target for 2029 is about a tenth of the size Rowe described, and the 2,000 mark sits several years further out. Other vendors publish other plans, and algorithm improvements keep lowering the resource estimates for specific molecules, so the comparison does not settle the date. It does show why a pharmaceutical computing chief can treat quantum chemistry as a 2030s capability while still funding readiness today.
Quantum pillar: computing. Technology readiness: TRL 4 of 9. The hybrid workflow for the light-activated drug has run on IBM quantum hardware as a research demonstration, many steps before it could shape which drug candidates enter a clinical program.
What Cleveland Clinic, Algorithmiq and IBM showed on up to 100 qubits, and what Wellcome Leap paid for on 16 April 2026
The Cleveland Clinic project worked on photodynamic therapy. As the US National Cancer Institute explains, a patient receives a photosensitizing drug that collects in the target tissue, and light at a specific wavelength then activates it to kill cells. Designing better photosensitizers depends on predicting how a molecule behaves once it absorbs light, its excited states, and those calculations are hard for classical methods. The Monitor looked at why this drug class is a good test for quantum computing in an earlier post on the two jobs a photosensitizer must do.
Wellcome Leap announced on 16 April 2026 that the Algorithmiq-led team had met the criteria for a $2 million prize in its $50 million Quantum for Bio program. That prize rewarded an experimental realization on a quantum computer with more than 50 qubits, at a program depth of roughly a thousand to ten thousand operations, with a clear path toward scale. A separate $5 million grand prize was reserved for demonstrated quantum advantage over leading classical methods, and the announcement awarded only the $2 million prize. Algorithmiq's own release says the workflow used up to 100 qubits on IBM hardware to simulate the activation pathway of a photosensitizer drug currently in Phase II clinical trials, and fed the resulting data into an active-learning AI loop that proposes new candidates. These are physical qubits on today's hardware, without the error correction that defines a logical qubit.
The team was led by Sabrina Maniscalco of Algorithmiq, with Ivano Tavernelli of IBM and Vijay Krishna of Cleveland Clinic as co-principal investigators. Krishna described the hospital's role as defining the biomedical problem and anchoring it in a clinically relevant use case. That division of labor matters for hospitals reading this: the clinical partner chose the question, and the software and hardware partners built the computation around it.
What Lara Jehi asked of hospitals, and what a research office can do with it in 2026 and 2027
Jehi's argument was historical. "In life sciences, in particular, we've always been late to the game when technology is being developed," she said, which leaves hospitals with tools that were not built for their problems. Cleveland Clinic is now exploring drug discovery, chemistry and biomarker discovery so its researchers can move onto more capable hardware as it arrives. "We have to figure it out now, at least to be ready for when that hardware is ready," she said.
Rowe's and Jehi's positions fit together more closely than the headlines suggest. Both expect useful chemistry to need much larger machines than exist today. Both want the people and the problem definitions ready before those machines arrive. Quentir has described this shift before as quantum drug discovery entering a workflow phase, where the value in the near term sits in the pipeline around the quantum step.
For a hospital research office, three questions follow from the October talks. Which of its own clinical problems involve molecular or excited-state chemistry that classical methods handle poorly? Which partners would let its clinicians define the problem, as Krishna's team did? And what would it need to see, in a published comparison against leading classical methods, before any result touched a drug or device decision? The $2 million award recognized an experimental demonstration on quantum hardware, and it does not establish quantum advantage; Wellcome Leap's announcement awarded no grand prize for that result.
For buyers, the practical reading is a timetable. A demonstration on up to 100 physical qubits exists today, an error-corrected machine of about 200 logical qubits is one major vendor's stated target for 2029, and the 2,000 logical qubits that a GSK computing lead associates with useful chemistry sit around 2030 in his telling and in 2033 or later on IBM's. Expertise built in the meantime is the part of the investment that does not depend on which date proves right.
Sources
Primary source: Computer Weekly, report of 19 October 2026 on the City Quantum & AI Summit, quoting Anthony Rowe of GSK and Lara Jehi of Cleveland Clinic. Also drawn on: Wellcome Leap's Quantum for Bio prize announcement of 16 April 2026, Algorithmiq's release on the prize (Sabrina Maniscalco, Ivano Tavernelli, Vijay Krishna), IBM's quantum roadmap and its June 2025 fault-tolerance post, the US National Cancer Institute's page on photodynamic therapy, and earlier Quentir coverage. The readiness rating is the Monitor's editorial assessment.