Danish scientists moonlight with a printer-sized quantum computer to design peptides — first practical drug-discovery output
TL;DR
A DTU team on weekend side-work paired generative AI with an ORCA Computing printer-sized quantum machine to design peptides — biggest gains in data-scarce fields.
A DTU (Technical University of Denmark) team using weekend hours and leftover project funds paired generative AI with a printer-sized quantum machine from UK's ORCA Computing to design new peptides that bind to specific human proteins. The experiment showed that the hybrid method performs best in domains where training data is scarce, producing more successful peptide sequences than classical computing alone.
The work is led by Professor Timothy Patrick Jenkins and aims to accelerate personalized immunotherapies and vaccine development, and to improve drug efficacy for under-researched populations in Asia and Africa — training data skews toward Western populations, and quantum-assisted few-shot learning fills that gap. Jenkins admits he was a "quantum skeptic" until this result changed his mind.
The path here runs alongside mainstream quantum computing, not down it. IBM, Google, and IonQ are chasing hundreds to thousands of qubits and general-purpose compute; ORCA runs a specialized photonic quantum machine — small, low-power, and good only at specific machine-learning workloads. This result is the first time a specialized small-scale quantum computer has delivered usable output in drug discovery — not a demo, but a real edge in successful peptide count.
Next steps: scale the pipeline to larger proteins and more advanced models, and explore designing synthetic antivenom for snake bites — a scenario where training data is inherently scarce (a few hundred venom samples per year, not tens of thousands), which is quantum-assisted learning's sweet spot.
The "weekend side-project" framing reflects that quantum-compute cost structures have reached a threshold: a desktop-scale photonic quantum machine plus a PhD student's off-hours can now produce top-journal results.
via WIRED
The work is led by Professor Timothy Patrick Jenkins and aims to accelerate personalized immunotherapies and vaccine development, and to improve drug efficacy for under-researched populations in Asia and Africa — training data skews toward Western populations, and quantum-assisted few-shot learning fills that gap. Jenkins admits he was a "quantum skeptic" until this result changed his mind.
The path here runs alongside mainstream quantum computing, not down it. IBM, Google, and IonQ are chasing hundreds to thousands of qubits and general-purpose compute; ORCA runs a specialized photonic quantum machine — small, low-power, and good only at specific machine-learning workloads. This result is the first time a specialized small-scale quantum computer has delivered usable output in drug discovery — not a demo, but a real edge in successful peptide count.
Next steps: scale the pipeline to larger proteins and more advanced models, and explore designing synthetic antivenom for snake bites — a scenario where training data is inherently scarce (a few hundred venom samples per year, not tens of thousands), which is quantum-assisted learning's sweet spot.
The "weekend side-project" framing reflects that quantum-compute cost structures have reached a threshold: a desktop-scale photonic quantum machine plus a PhD student's off-hours can now produce top-journal results.
via WIRED
