Industries / Technology / Quantum computing
Software and algorithms
This is the work of finding problems a quantum computer can solve faster, writing the programs that run on it, and testing whether ordinary computers can match the results.
7
Milestones achieved, from 1994 to 2023.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1994
Shor's factoring algorithm
Peter Shor at AT&T showed that a quantum computer could factor whole numbers and find discrete logarithms in a number of steps that grows only polynomially with the number of digits. Both problems are believed hard for ordinary computers and underpin public-key encryption such as RSA. He presented it at the FOCS conference in November 1994.
Credit
Peter Shor, AT&T Research
1996
Grover's search
Lov Grover at Bell Labs showed that a quantum computer can find one marked item among N in about √N steps, where an ordinary search needs to look at N/2 items for an even chance, and that no quantum method can do much better.
Credit
Lov Grover, Bell Labs
2001
Fifteen factored
Lieven Vandersypen, Isaac Chuang and colleagues at IBM Almaden and Stanford ran the simplest case of Shor's algorithm, factoring 15 into 3 and 5, using seven atomic nuclei in a molecule as qubits, controlled by nuclear magnetic resonance. They noted that the method's scalability "is not implied by the present work".
Lessonin 2013 John Smolin, Graeme Smith and Alex Vargo showed that a "compiled" circuit, built using the answer, can appear to factor any such number with two qubits, and in 2025 Peter Gutmann and Stephan Neuhaus matched the published quantum factoring records with a 1981 VIC-20 home computer, an abacus and a dog; the size of a number factored says little unless the circuit was built without knowing the factors.
Credit
Lieven Vandersypen, Matthias Steffen, Isaac Chuang and colleagues
2013
Hybrid chemistry runs
Alberto Peruzzo, Jarrod McClean, Jeremy O'Brien and colleagues introduced the variational quantum eigensolver, or VQE, in which an ordinary computer tunes a short quantum circuit until it finds a molecule's lowest energy. On a small photonic processor they computed the energy of the helium hydride ion (HeH+) to chemical accuracy. Short circuits made chemistry possible on noisy machines, and the method spread: the silicon processor in the 2021 hardware entry above ran VQE.
LessonThibaud Louvet, Thomas Ayral and Xavier Waintal argue that noise harms VQE so much that useful chemistry would need fault-tolerant machines, even with error mitigation.
Credit
Alberto Peruzzo, Jarrod McClean, Alán Aspuru-Guzik, Jeremy O'Brien and colleagues
2017
Open programming tools
IBM released Qiskit, open-source software for writing quantum programs in Python, with Jay Gambetta's first code on GitHub on 3 March 2017 and version 0.1 on 7 March. It now runs on machines from many providers: Amazon Braket accepts Qiskit programs for Rigetti's Cepheus processor.
Credit
Jay Gambetta and the IBM Qiskit team
2018
A speedup taken back
Ewin Tang found an ordinary algorithm for recommendation systems, the software that suggests products, that runs only polynomially slower than a quantum algorithm by Kerenidis and Prakash. That quantum algorithm had been seen as one of the strongest candidates for an exponential speedup in quantum machine learning.
Lessona quantum speedup has to be measured against the best ordinary algorithm given the same access to the data, and that algorithm may not exist yet.
Credit
Ewin Tang
2023
Utility claimed, then matched
IBM simulated a model of interacting magnets, called a kicked Ising model, on 127 qubits and used error mitigation to estimate results that it presented in Nature, on 14 June 2023, as evidence that quantum computers are useful before error correction. Within two weeks, Joseph Tindall and colleagues reproduced it more accurately with tensor networks, a compact way to store quantum states on ordinary computers, and Tomislav Begušić and Garnet Chan did it on a single laptop core faster than the quantum run.
Lessonclassical methods improve in response to each claim, so a claim of advantage needs time and public challenge before it counts.
Credit
IBM Quantum; Joseph Tindall, Matt Fishman, Miles Stoudenmire and Dries Sels; Tomislav Begušić and Garnet Chan
Watch for
Verifiable advantage by echoes
Google says its Quantum Echoes algorithm, built on a quantity it calls an out-of-order time correlator, ran on Willow 13,000 times faster than the best classical algorithm on one of the fastest supercomputers, and that it can be used to learn the structure of molecules. It calls this, in a Nature paper, the first verifiable quantum advantage.
Achieved whena group outside Google confirms the values on other hardware, and no published classical method has matched the speed after outside teams have tried.
Claimed by
Google Quantum AI, October 2025
Advantage on irregular materials
Algorithmiq and IBM say a simulation of a quantum material with irregular regions, run on IBM machines, has outlasted classical challenges: eight months after they posted it on the Quantum Advantage Tracker, no classical method had reliably reproduced its results across the full range studied. Algorithmiq also released monoprop, its best classical method, so others can test the claim.
Achieved whena group not involved confirms the quantum results are right, for example on a different quantum computer or with a bounded classical calculation on part of the range, and no published classical method matches them.
Claimed by
Algorithmiq and IBM, July 2026
Open
01
An honest factoring test
Run Shor's algorithm for 21 and 35 on rented hardware with circuits built by a program that never sees the factors, following the fairness criteria Gutmann and Neuhaus set out. Publish the circuits, raw counts and success rate against what random outputs would give, whether the attempt succeeds or fails.
The briefCredit · unclaimed02
Chemistry claims against a laptop
A public table of molecular-energy calculations run on quantum hardware, each rerun with standard classical chemistry methods on a laptop, listing the molecule, basis set, qubits used, the quantum error against the exact answer, the classical error and the laptop run time.
The briefCredit · unclaimed
Open 01 · the brief
An honest factoring test
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
Run Shor's algorithm for 21 and 35 on rented hardware with circuits built by a program that never sees the factors, following the fairness criteria Gutmann and Neuhaus set out. Publish the circuits, raw counts and success rate against what random outputs would give, whether the attempt succeeds or fails.
Why it is still open
A 2024 paper states that "all current machines lack a sufficient number of qubits and acceptable decoherence times to run a full non-compiled version of Shor's algorithm, even for small numbers". The 2019 attempt on IBM hardware for 15, 21 and 35 used compiled circuits, and Gutmann and Neuhaus matched the published records with a 1981 home computer. Machines rented in 2026 are larger and cleaner, and a November 2026 search found no report of the test on them.
Built with what exists
Qiskit and its simulator to check circuits before running them; IBM's Nighthawk r2, and a trapped-ion machine rented through Amazon Braket or Azure Quantum; Quantinuum's pytket compiler; a short analysis script for the period-finding step.
First step
Write the uncompiled circuit for 21 in Qiskit, simulate it, and count its two-qubit gates after compiling it for Nighthawk r2.
Done looks like
For 21 and 35, circuits generated without the factors, at least 10,000 shots each on two different machines, the share of runs that yield a correct period set against the random baseline, and a second person reaching the same numbers from the raw counts.
Open 02 · the brief
Chemistry claims against a laptop
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public table of molecular-energy calculations run on quantum hardware, each rerun with standard classical chemistry methods on a laptop, listing the molecule, basis set, qubits used, the quantum error against the exact answer, the classical error and the laptop run time.
Why it is still open
Louvet, Ayral and Waintal argue on theoretical grounds that noisy hardware cannot do useful chemistry with VQE, but a November 2026 search found reviews of the methods and no table that reruns each hardware result classically.
Built with what exists
PySCF, open-source quantum chemistry software; Qiskit to rebuild the published circuits in simulation; an AI research agent to collect papers and draft rows, each checked by a person.
First step
Rerun the HeH+ energy from the 2013 VQE paper in PySCF and record the error and run time.
Done looks like
30 hardware results, each with its classical rerun and timing; a second person reproduces 10 of them; each paper's authors invited to comment before publication.