XPRIZE announces finalists in quantum applications | Keryc
Today the seven finalists of the XPRIZE Quantum Applications are announced, a global 3-year, $5 million challenge backed by Google Quantum AI, Google.org and GESDA. From 133 proposals, these seven initiatives aim for quantum algorithms that outperform the best classical methods on concrete problems tied to real goals, like some of the UN Sustainable Development Goals.
Why does this matter now? Because moving from theoretical demos to verifiable advantages in practical applications is exactly the bottleneck the quantum community needs to speed up. The XPRIZE is meant to close that gap.
What XPRIZE is looking for and the technical context
The contest rests on a five-stage framework for application development, and it now pushes the middle stages where real advantages are defined and verified. Two key stages are:
Find the right instances (Stage II): locate concrete, verifiable problems where a quantum algorithm can beat the best classical alternative — not just in theory but on useful instances.
Establish real-world advantage (Stage III): link a mathematical advantage to an industrial or scientific use case that delivers tangible value.
Google Quantum AI brings relevant advances: the Willow chip, progress in error correction, and the Quantum Echoes algorithm, which showed the first demonstration of verifiable quantum advantage. But seeing an idea on paper and proving it against an efficient classical pipeline are two different things.
The 7 finalists
Below is a technical and practical summary of each finalist, their approach and the proposed impact.
Calbee Quantum (USA)
Team lead: Garnet Chan, Caltech.
Approach: a new way to simulate materials that gets a speedup in system size even with approximate accuracy.
Impact: practical simulations for semiconductors and optoelectronics.
Gibbs Samplers (Hungary)
Team lead: András Gilyén, HUN-REN.
Approach: a quantum algorithm to simulate thermalization of quantum systems at finite and low temperatures.
Impact: reduce the search space of candidates in materials discovery.
Phasecraft - Materials Team (UK)
Team lead: Toby Cubitt, Phasecraft.
Approach: use quantum simulations to improve classical quantum chemistry methods.
Impact: speed up discovery of materials for clean energy: batteries, solar cells, carbon capture.
The QuMIT (USA)
Team lead: Alexander Schmidhuber, MIT.
Approach: an algorithm that speeds up detecting communities in hypergraphs, a problem in applied computing.
Impact: improve analysis of protein–protein interactions for polygenic diseases and targeted therapies.
Xanadu (Canada)
Team lead: Juan Miguel Arrazola, Xanadu.
Approach: new representation and algorithm to simulate the time evolution of certain molecular processes.
Impact: support design of organic solar cells and applications in photodynamic therapy.
Q4Proteins (Switzerland)
Team lead: Markus Reiher, ETH Zurich.
Approach: a detailed framework combining quantum simulations with classical machine learning.
Impact: a first-principles simulation pipeline for drug discovery and study of biomolecular condensates.
QuantumForGraphproblem (USA)
Team lead: Jianqiang Li, Rice University.
Approach: a quantum algorithm to solve linear systems without the problematic dependence on the condition number that affected earlier methods.
Impact: enable quantum-advantaged applications across domains that need fast linear-system solutions.
Phase II: what they'll measure and why it's technical
Phase II focuses on rigorous performance evaluations. What metrics really matter?
Comparison against the best classical techniques: running against a weak baseline isn't enough.
Scalable speedup: measure how the advantage grows with problem size — for example, the relation between qubit count and simulable system size.
Latency and throughput: wall-clock time, and how many runs are needed to get statistically significant data.
Physical resources: qubit count, circuit depth, error rates and error-correction needs (overhead in qubits and time).
Robustness and reproducibility: stability under noise, and whether the advantage is verifiable in different labs.
Total compute cost and deployment feasibility: from state preparation to classical postprocessing.
They'll also compute detailed resource estimates: how much calibration, how many logical gates, and when a pipeline is feasible on real hardware versus ideal simulation. The big technical challenge is closing the gap between a proof of concept and an end-to-end advantage in industrial applications.
Implications for industry, science and realistic timelines
What does this mean for you if you work in materials, pharma or optimization?
For materials and energy: quantum simulations promise to explore chemical spaces that are currently inaccessible, reducing the number of experimental iterations for batteries or solar cells.
For biomedicine: more accurate interaction models can speed up drug hypotheses, though the full pipeline (simulation, ML, experimental validation) remains complex.
For optimization and data science: algorithms that remove harmful dependencies like the condition number open new practical applications in finance, logistics and modeling.
Does this mean classical methods will be replaced tomorrow? No. But these initiatives are measurable steps toward applicable advantages. Verifiable and reproducible advantage is what will convince industry to invest in real deployments.
What’s next
The finalists move to Phase II, where they'll face demanding benchmarks and the need to standardize tests against classical methods. One million dollars is being shared now; the remaining $4 million will be decided in March 2027, including $3 million for the grand prize. Teams not selected can return via a wildcard round in 2026.
The technical lesson is clear: quantum progress isn't just hardware or algorithms in the abstract. It's about combining clever representations, resource analysis and rigorous tests against the real world. If you're interested in the intersection of physics, algorithms and industrial applications, these seven finalists are the projects to follow.
