Which Problems Will Quantum Computers Solve—and When?

The 21st-century fervor astir building nan first industrial-scale quantum computer, pioneering theoretical physicist Peter Zoller says, is akin to nan 20th-century obsession pinch becoming nan first to conquer Mount Everest. “When you’re climbing, you look astir worrying, ‘Who is number one?’” he says. “When you scope nan top, that’s erstwhile you inquire yourself, ‘Why nan hellhole did we really do this?’”

In 1995 Zoller and Ignacio Cirac, past a postdoctoral interrogator successful Zoller’s group astatine nan University of Colorado Boulder, projected nan first realistic blueprints for a quantum computer. Their thought was to usage trapped ions arsenic “qubits”—the quantum balanced of integer bits, capable to beryllium successful a superposition that simultaneously represents 0, 1 and each positions successful between. More than a decade earlier physicists Paul Benioff and Richard Feynman had independently suggested that machines harnessing nan quantum realm’s weirdness could, successful theory, outperform classical computers astatine immoderate tasks.

Today teams astir nan world are processing ever bigger quantum processors utilizing qubits made from ions, neutral atoms, superconducting loops, and more. IBM and Berkeley, Calif.–based institution Atom Computing presently lead nan complaint pinch quantum computers hosting much than 1,000 qubits, and past twelvemonth a investigation group astatine nan California Institute of Technology reported that it had built a record-breaking array of much than 6,000 qubits.

On supporting subject journalism

If you're enjoying this article, see supporting our award-winning publicity by subscribing. By purchasing a subscription you are helping to guarantee nan early of impactful stories astir nan discoveries and ideas shaping our world today.

“It’s an breathtaking clip because group are fielding quantum computers pinch hundreds and thousands of qubits,” says Nobel Prize–winning quantum physicist John Martinis, a professor emeritus astatine nan University of California, Santa Barbara, and co-founder of quantum hardware institution Qolab.

In 2019 Google researchers, led by Martinis, reported that their 53-qubit processor, Sycamore, had go nan first to execute “quantum advantage,” performing a calculation successful 200 seconds that they estimated would return nan champion classical supercomputers astir 10,000 years to solve. That supercomputer number was later disputed—IBM based on that its champion classical machine could really execute nan task successful conscionable 2 and half days—but moreover if it held up, nan calculation was of only world liking arsenic a impervious of principle. “Google’s 2019 objection of quantum [advantage] was an important milestone, but galore group would opportunity that it did not yet represent a breakthrough connected a problem of wide applicable significance,” notes quantum physicist Kihwan Kim, now astatine nan Institute for Basic Science successful South Korea.

“We tin now simulate things for illustration superconductivity, artificial photosynthesis and mini supplier designs.” —Michelle Simmons, Silicon Quantum Computing

Experts work together that to tackle useful problems that dishonesty beyond nan scope of moreover nan champion imaginable classical supercomputers, we request qubit numbers to jump significantly, perchance to a cardinal aliases more. In addition, quantum physicists will request to technologist robust qubits that support their quantum properties for a longer time, and they’ll person to find ways to hole errors introduced during calculations. In March a multi-institutional group of researchers reported that IBM’s superconducting Heron processors could accurately foretell nan results of neutron-scattering experiments that measured nan building of a circumstantial antiferromagnetic crystal to an unprecedented scale, utilizing 50 qubits aliases fewer; nan physicists noted, however, that classical computers could execute nan aforesaid feat faster and much accurately.

So what will quantum computers really beryllium bully for—and when? Experts opportunity we are still years distant from quantum computers capable to grip applicable applications that classical computers cannot, which mightiness see breaking communal data-encryption schemes, simulating quantum processes for basal physics, and designing amended narcotics and materials. That said, Martinis—an master connected scaling up quantum hardware—notes location are nary guarantees that million-qubit computers will ever beryllium created. “The proof,” he says, “will beryllium building them and seeing that they work.”

Cryptography

The astir notorious committedness of quantum computers has been that they will 1 time break RSA encryption, a long-standing protocol utilized worldwide to unafraid slope transfers, cryptocurrencies and integer communication. That time whitethorn travel amazingly soon. It was agelong thought that cracking encryption would require a processor pinch astatine slightest a cardinal qubits. But successful February a squad from Iceberg Quantum successful Sydney, Australia, dramatically reduced that estimate, calculating that pinch observant optimization and correction correction, hackers mightiness request less than 100,000 qubits for nan feat. In March, Google announced a caller committedness to migrating its systems by 2029 to protect them from quantum hacking.

Although nan Iceberg claims person yet to beryllium peer-reviewed, they are reliable and person caused a stir, says Artur Ekert, a cryptography master astatine nan University of Oxford. “Many deliberation that nan threat to encryption from quantum computers is conscionable mumbo jumbo—and I person besides been skeptical—but it could conscionable return a fewer much papers for illustration this 1 to make nan conception of breaking RSA relevant,” he says. Martinis’s sentiment has besides shifted successful caller years. “If you are worried astir RSA encryption—as you should be—I would opportunity it mightiness beryllium surgery successful 5 to 10 years,” he says.

RSA encryption leverages nan truth that it’s easy to create a concealed cardinal by multiplying 2 ample premier numbers together but efficaciously intolerable for immoderate classical machine to find nan cardinal by efficiently factoring it backmost into those constituent primes. A classical machine could trial successive numbers sequentially, remembering each worth and looking for a pattern, but this attack is infeasible pinch ample numbers, Ekert explains. Modern classical algorithms usage different methods but stay inefficient because nan execution clip increases exponentially pinch nan size of nan number to beryllium factored.

Jen Christiansen; Source: The Quantum Index Report 2025, by Jonathan Ruane et al.; M.I.T. Initiative connected nan Digital Economy, Massachusetts Institute of Technology, May 2025 (data)

The quantum world, however, is not truthful constrained. Qubits tin return connected aggregate values simultaneously and go entangled pinch 1 another, amplifying their power. “Essentially it tin screen each imaginable computational paths astatine nan aforesaid time,” Ekert says. In 1994 theoretical machine intelligence Peter Shor, now astatine nan Massachusetts Institute of Technology, projected that a hypothetical quantum machine could usage this spot to ace RSA encryption. If and erstwhile 1 does, it will astir apt usage nan algorithm Shor developed.

There are proposals for quantum-resistant cryptographic algorithms; nan U.S. National Institute of Standards and Technology (NIST) published 3 specified schemes successful 2024. Zoller thinks specified activity mostly defuses nan threat because it suggests nan world tin move distant from RSA encryption earlier quantum hackers arrive. “Shor’s algorithm whitethorn yet beryllium remembered arsenic a landmark technological accomplishment of awesome humanities value for inspiring nan improvement of quantum computers, arsenic overmuch arsenic for its implications for breaking encryption,” he says.

Ekert feels little reassured. Last year, he notes, a machine intelligence successful China proposed—it turns retired incorrectly—a quantum algorithm tin of breaking NIST’s apical candidate, which is called lattice-based encryption. “It took nan brainpower of nan full quantum cryptography organization [more than a week] to find a mistake, showing you really adjacent those things are,” Ekert says. “Maybe adjacent clip it will beryllium correct.”

Fundamental Physics

One domain wherever quantum processors are already seeing occurrence is successful modeling particle interactions to lick mysteries astatine nan bosom of basal physics. “It goes backmost to Feynman, who articulated that you can’t really understand really quality useful unless you build it astatine nan aforesaid magnitude scale,” says quantum physicist and worldly intelligence Michelle Simmons, laminitis and CEO of Silicon Quantum Computing (SQC) successful Sydney.

Simulating interactions of aggregate particles quickly becomes intolerable pinch a classical computer, explains Daniel González-Cuadra, a quantum physicist astatine nan Institute for Theoretical Physics successful Austria. “The accusation that it takes to picture nan authorities of these systems grows exponentially pinch nan size of nan system, and astatine immoderate constituent you conscionable don’t person capable memory,” he says. Capturing each nan complexities requires an arsenic analyzable quantum machine.

Groups astir nan world are making advancement connected this front. One large area of attraction is expanding nan “coherence” of qubits truthful they enactment successful superposition agelong capable to transportation retired their calculations. In 2021 Kim’s China-based squad demonstrated that trapped-ion qubits tin support coherence for much than an hour, which he says was “a very important benchmark for scaling up meaningful quantum simulations.”

Jen Christiansen; Source: “Simulation of Matter–Antimatter Creation connected Quantum Platforms,” by Michele Burrello, successful Nature, Vol. 642; June 12, 2025 (reference)

Last twelvemonth 2 teams independently published real-time quantum simulations of nan creation of matter and antimatter during a process called drawstring breaking. According to nan Standard Model of particle physics, pairs of powerfully interacting subatomic particles, specified arsenic quarks, behave arsenic though they are joined by an elastic string, “like a violin drawstring that vibrates,” explains quantum physicist Pedram Roushan of Google Quantum AI successful Santa Barbara. Roushan’s squad ran its simulation connected Google’s Sycamore chip, which uses superconducting loops arsenic qubits. The simulation showed really pulling 2 particles isolated increases nan string’s hostility until it yet snaps, releasing nan stored power by generating a caller brace of matter and antimatter particles. “These theoretical concepts were known since nan 1970s, but we were capable to visualize them and return a image of nan strings and their breaking,” Roushan says.

The Sycamore research was an illustration of a integer simulation, meaning it was performed connected a multipurpose spot pinch circuits of qubits designed to do galore different tasks. In contrast, González-Cuadra, Zoller and their colleagues worked pinch a squad astatine QuEra Computing successful Boston to create an analog simulator—a lattice of neutral-atom qubits specially built to simulate drawstring breaking. The 2 string-breaking simulations are among nan first to exemplary particle interactions successful 2 spatial dimensions, González-Cuadra says. “The physics is richer, truthful we could spot really these strings fluctuate,” he explains.

“We’re optimistic that we’re going to spot nan first applicable applications successful 5 years.” —Sergio Boixo, Google Quantum AI

These kinds of simulations won’t switch particle physics experiments. Instead they will thief physicists hone their theories and make testable predictions that tin beryllium checked astatine particle accelerators. And truthful acold they person simulated only elemental models that tin besides beryllium checked pinch classical computers. But González-Cuadra believes quantum simulators will commencement to surpass their classical counterparts successful a mates of years, marking an era of existent quantum advantage. This anticipation raises nan mobility of really physicists tin beryllium judge their quantum simulations are spitting retired reliable results. To reply it, past twelvemonth Zoller and his colleagues posted a preprint insubstantial connected arXiv.org describing a strategy for processing an analog quantum instrumentality that not only makes predictions but besides quantifies nan uncertainty successful those predictions. “If you inquire maine what nan large situation for quantum simulation is, it is nan frontier of verification,” Zoller says.

Materials Design

The dream, Zoller says, is for quantum simulators to displacement from a passive “discovery mode”—in which they are utilized to exemplary nature—to an “active creation mode” successful which quantum computers would spit retired recipes for synthesizing caller molecular structures pinch circumstantial desirable properties. The quantum engineering of caller molecules could lead to amended narcotics and to batteries that don’t usage costly, environmentally damaging commodities specified arsenic uncommon world elements. “These things return billions of dollars, truthful if you conscionable make thing moreover a fewer percent cheaper aliases a fewer percent better, past that’s really worthy it,” Martinis says. “It could beryllium a immense point not conscionable monetarily but for changing really things are built to make them much ecological.”

Room-temperature superconductivity is 1 high-priority target. Superconductivity, nan free travel of energy without resistance, typically requires a worldly to beryllium cooled to highly debased temperatures, which makes it impractical for galore applications. But definite materials grounds nan arena astatine higher temperatures, and immoderate researchers dream quantum engineering tin thief them find caller superconducting materials that don’t request immoderate cooling astatine all. “These systems dwell of 1023 particles, whereas classically we tin exemplary only astir 100 particles,” says Henrik Dreyer, a quantum physicist astatine Quantinuum successful Munich. Physicists would request to trim nan correction rates successful quantum processors to conscionable 1 successful a cardinal to make it possible; astatine nan infinitesimal nan champion chips are down to somewhat beneath 1 successful 1,000, Dreyer explains.

Dreyer and his colleagues person been performing integer simulations of cuprate superconductors utilizing Quantinuum’s Helios chip, which employs 98 trapped-ion qubits. In cautiously controlled laboratory conditions, shooting these materials pinch a laser tin very briefly—and surprisingly—create a superconducting authorities astatine a comparatively precocious temperature. “The first mobility is: Why?” Dreyer says. Last twelvemonth Quantinuum posted a preprint connected arXiv saying its two-dimensional simulation modeling nan worldly shows that nether laser fire, its electrons brace up—a information basal for travel without resistance. “The eventual mobility is,” arsenic posed by Dreyer, “Can we technologist it to do this astatine room somesthesia for a minute, an hour, 10 days, aliases more?”

Jen Christiansen; Source: “Superconducting Pairing Correlations connected a Trapped-ion Quantum Computer,” by Etienne Granet et al.; February 17, 2026 (arXiv:2511.02125v3) (reference)

Meanwhile Simmons and her SQC colleagues successful Australia person developed a simulation strategy called Quantum Twins—a 2D array of 15,000 clusters of phosphorus atoms embedded successful silicon—to create analogs of various materials. In February nan squad reported that nan level tin simulate nan modulation betwixt insulating behaviour and metallic conduction. “We tin now commencement to simulate things for illustration superconductivity, different artillery materials, artificial photosynthesis and mini supplier designs,” Simmons says.

Google Quantum AI’s Sergio Boixo notes that nan institution has collaborated pinch BASF connected artillery design, Sandia National Laboratories successful Albuquerque connected fusion energy, and German chemic institution Covestro connected pharmaceutical development. Last twelvemonth it implemented an algorithm for modeling molecular building connected Willow, Google’s 105-qubit superconducting processor, that tin beryllium utilized successful operation pinch atomic magnetic spectroscopy. The technique, which useful by bouncing signals onto qubits and efficaciously listening for their echoes, runs 13,000 times faster connected Willow than an balanced algorithm would connected nan champion classical supercomputer. One important facet of nan algorithm’s creation is that it allows results to beryllium corroborated by different quantum machine. “Quantum Echoes is nan world’s first quantum-verifiable algorithm pinch quantum advantage,” Boixo says. “We’re optimistic that we’re going to spot nan first applicable applications successful 5 years.”

Quantum AI

If you really want to make hype, harvester nan connection “quantum” pinch “AI,” jokes Jacob Biamonte, an master connected quantum instrumentality learning astatine ÉTS Montreal. Indeed, arsenic quantum processors get bigger, immoderate physicists are focusing connected utilizing them to boost nan capacity and power ratio of classical artificial intelligence.

Last twelvemonth SQC launched Watermelon, a quantum-enhanced AI processor, to thief velocity up instrumentality learning. Classical AI systems are already adept astatine uncovering patterns successful immense datasets, which makes them peculiarly useful for optimizing communications and power networks, for instance. SQC’s quantum method builds connected classical reservoir computing, a method for taking input information points and transforming them onto a higher-dimensional neural network, making it easier to find patterns. In 2017 scientists successful Japan predicted that nan classical nodes of nan neural web could beryllium replaced by a smaller number of qubits taxable to quantum interference. “The advantage of having a quantum reservoir is that you get an exponential summation successful dimensionality,” Simmons says, enabling a quantum reservoir to execute nan aforesaid training results arsenic a classical reservoir but perchance faster and utilizing less resources.

Jen Christiansen; Source: “Quantum Reservoir Computing Implementation connected Coherently Coupled Quantum Oscillators,” by Julien Dudas et al., successful npj Quantum Information, Vol. 9; July 7, 2023 (reference)

Watermelon’s first commercialized trial—in collaboration pinch Australian telecommunications institution Telstra—has shown promising results. Telstra already uses AI to show latency and bandwidth patterns connected its networks. It takes astir 3 weeks to train nan company’s models utilizing modular classical methods. With Watermelon’s help, Telstra achieved nan aforesaid training results successful conscionable 2 days. “In nan expansive strategy of nan world, that is rather important because astatine nan moment, information centers are very powerfulness hungry,” Simmons says, noting that akin optimizations could beryllium quickly rolled retired to different energy-intensive tasks, specified arsenic training AIs for image recognition, fraud discovery and marketplace prediction. “I consciousness for illustration I’m successful this freight train that’s going astatine superhigh speeds,” she says.

Ekert, however, remains cautious astir nan longer-term benefits of utilizing quantum AI processors to analyse classical datasets. “Turning classical information into a quantum shape is terribly inefficient,” he says. Where quantum computing and instrumentality learning are already being mixed astir helpfully, Ekert argues, is successful physicists’ usage of classical AI to creation quantum error-correcting codes and amended quantum hardware. Last year, for example, Finnish institution QMill launched a classical AI work for compressing quantum circuits, reducing nan number of gates needed for cognition by 20 to 50 percent. Biamonte besides thinks nan existent imagination is excessively small. “If nan extremity is to usage quantum computers to do instrumentality learning for classical data, it doesn’t moreover make sense, because classical instrumentality learning is already truthful good,” he says.

If quantum processors could 1 time beryllium utilized to analyse quantum information directly, however, that would beryllium a crippled changer. “There should beryllium these awesome patterns that classical computers cannot observe because location are conscionable excessively galore information for their memory,” Biamonte says. A quantum AI could riff connected nan molecular building of an existing patented drug, for instance, to make aggregate different configurations pinch nan aforesaid benefits. It could past measure those molecules to spot whether they could beryllium synthesized and patented earlier a institution committed costs to trying to make them. “That’s nan breathtaking early that doesn’t beryllium yet,” Biamonte says.