If you’ve searched for the latest breakthroughs in quantum computing 2024, you’ve probably run into two very different stories. One says quantum computers are about to change everything. The other says nothing useful has happened yet.
Neither one gives you the full picture, and that gap is confusing if you’re trying to make a real decision, whether that’s an investment, a research direction, or just an honest answer to “is this real?”
This article walks through what actually happened in 2024, chip by chip, result by result, and holds every claim next to its actual limits.
A Quick Primer Before We Start
A classical computer stores information as a bit, either 0 or 1. A qubit can hold a mix of both states at once, a property called superposition. Two or more qubits can also become linked through entanglement, so measuring one instantly affects the other, even at a distance.
This is where physical qubits come in. A physical qubit is a single hardware unit, built from a superconducting circuit, a trapped ion, or another quantum system. On its own, it’s noisy and error prone. While a logical qubit is built by combining many physical qubits into one error corrected unit that behaves more reliably.
A fault-tolerant quantum computer is one where logical qubits stay accurate long enough, and in large enough numbers, to run real programs. Most of the news in 2024 sits somewhere between these three stages, and knowing the difference helps you read the headlines correctly.
The Real Progress: Breakthroughs in Quantum Computing 2024
Error Correction Finally Crosses a Threshold
The single biggest story among the latest breakthroughs in quantum computing 2024 came from Google Quantum AI. On December 9, 2024, the company unveiled Willow, a 105-qubit superconducting chip built in its Santa Barbara fabrication facility.
The milestone came from an experiment where Willow stored a single qubit of information across a 2D grid of physical data qubits, using grid sizes of 3×3, 5×5, and 7×7.
What made this significant wasn’t the qubit count. It was the direction of the error rate. Google reported that Willow was the first processor where error corrected qubits get exponentially better as they get bigger, with the encoded error rate dropping by roughly half each time the lattice grew from 3×3 to 5×5 to 7×7. That behavior is what researchers call below threshold error correction, and reaching it has been an open problem since Peter Shor first proposed quantum error correction in 1995.
Notably, the distance-seven logical qubit on Willow outlived every one of the 101 physical qubits it was built from — a long-sought “break-even” point for error correction. Google also ran a random circuit sampling benchmark on Willow, and the chip completed it in about five minutes, a task that would take today’s fastest supercomputer somewhere around 10 septillion years.
A few months earlier, in April 2024, Microsoft and Quantinuum published a separate but related result. Using a qubit virtualization system on Quantinuum’s 32-qubit H2 trapped-ion processor, the two companies created four logical qubits out of 30 physical qubits, reaching a circuit error rate of roughly one in 100,000 runs, an 800 times improvement over the physical qubit error rate of about eight in 1,000.
The team also ran more than 14,000 individual instances of a quantum circuit without a single uncorrected error. Microsoft described the result as a shift from Level 1 Foundational quantum computing to Level 2 Resilient quantum computing, the point where logical qubits start meaningfully outperforming the physical hardware underneath them.
A third team pushed logical qubit count in a different direction entirely. Mikhail Lukin and Dolev Bluvstein, working with colleagues at Harvard, MIT, and QuEra Computing, built a neutral atom processor holding 48 logical qubits at once, far more than earlier systems had managed to integrate together. Physics World named this work, together with Google’s Willow result, joint Physics Breakthrough of the Year for 2024, calling error correction the field’s defining challenge and praising both teams for tackling it through very different hardware.
Topological Qubits Take an Early Step Forward
Separately from the error correction results, Quantinuum worked with researchers at Harvard, Caltech, and the University of Chicago on a different approach to reliability, encoding information in the topology of a quantum system rather than correcting errors after the fact.
In November 2024, the team reported what they called the first experimentally demonstrated topological qubit, built on Quantinuum’s H2 ion-trap processor using a Z3 toric code and qutrits, three-level systems rather than standard two-level qubits.
The idea traces back to non-Abelian anyons, exotic particle-like states that Quantinuum, Harvard, and Caltech had already created and braided on the same H2 machine back in 2023. This line of research is still early.
The demonstrations run on a small number of qubits, and a full topological quantum computer is a long way off. The appeal is real, though. A topological qubit could, in theory, need far fewer physical qubits per logical qubit than surface code error correction, which is one reason several labs are chasing it in parallel.
Hardware Diversity Grows
2024 also made clear that quantum computing 2024 isn’t a one-architecture race. IBM’s Heron processor, a 133-qubit superconducting chip with tunable couplers to cut down crosstalk between qubits, became the foundation of IBM’s modular Quantum System Two. IBM reported a three to five times performance improvement over its earlier 127-qubit Eagle chip, with Heron able to run close to 1,800 gate operations within a qubit’s coherence window. Alongside the hardware, IBM shipped Qiskit 1.0, a rebuilt version of its open source quantum programming toolkit built for speed and stability.
Trapped-ion computing kept advancing through Quantinuum’s H2 system, the same machine behind the error correction and topological qubit work above. Neutral atom computing, from companies like QuEra and Pasqal, uses laser tweezers to hold individual atoms in place, and its main advantage is reconfigurable qubit layouts, letting researchers rearrange atoms mid-computation instead of being locked into a fixed chip design. Photonic quantum computing, which encodes information in particles of light, kept gaining attention for a different reason. It doesn’t need the extreme cryogenic cooling that superconducting qubits do, and can, in principle, run closer to room temperature. NVIDIA also expanded CUDA-Q, a platform meant to bridge classical GPU supercomputing with quantum hardware from multiple vendors, part of a broader industry move toward hybrid quantum-classical computing rather than treating quantum machines as a standalone replacement for classical systems.
Real-World Pilots Start Showing Up
Chips and papers are one thing. Actual pilot projects tell you whether any of this is close to useful. A few stood out in 2024.
IBM and Moderna published a case study applying quantum computing to mRNA secondary structure prediction, a step in designing mRNA-based medicines. Using 80 qubits on an IBM Quantum Heron chip, the team ran a CVaR-based variational quantum algorithm to predict the secondary structure of a 60 nucleotide mRNA sequence, the longest sequence simulated on quantum hardware at that point. The predictions matched results from classical solvers like CPLEX, even with fairly basic error mitigation, though the researchers were careful to note that scaling the method past 100 qubits remains an open question.
Microsoft’s Azure Quantum Elements platform, which runs on classical and AI-assisted computing today rather than quantum hardware, paired with the Department of Energy’s Pacific Northwest National Laboratory to search for new battery materials. The team screened 32 million candidate inorganic materials down to 18 promising options in about 80 hours, and the resulting solid-state electrolyte used up to 70% less lithium than materials found through traditional lab methods. Microsoft has said the platform is being built with an eye toward eventual quantum computing integration once the hardware catches up.
On the security side, JPMorgan Chase worked with Toshiba and Ciena to expand quantum key distribution across its fiber network, building on earlier joint work and presenting new results at the 2024 Optical Fiber Communications conference. The goal is a quantum-secured, crypto-agile network built to protect financial data against future quantum-era attacks, not just today’s classical ones.
The Real Challenges: An Honest Reality Check
None of the results above mean fault-tolerant quantum computing has arrived. Here is how the progress lines up against what’s still missing.
| Breakthrough | What Improved | What’s Still Limited |
| Willow error correction | Below threshold scaling confirmed for the first time | Still 105 physical qubits total, far short of the millions needed for large-scale fault tolerance |
| Microsoft and Quantinuum logical qubits | 800x lower error rate than physical qubits | Only 4 logical qubits demonstrated, on a 32-qubit machine |
| Harvard, MIT, and QuEra neutral atom processor | 48 logical qubits integrated at once | Still a research-scale demonstration, not a production system |
| Quantinuum topological qubit | A naturally more error-resistant qubit design | Extremely small-scale proof of concept, built on qutrits |
| Industry pilots (Moderna, JPMorgan, Azure Quantum Elements) | Real use cases in chemistry, biology, and network security | No proven quantum advantage yet over classical methods in these specific pilots |
Beyond the table, a handful of structural problems remain, and they explain why fault-tolerant quantum computing is still measured in years, not months.
Scaling is the biggest one. Useful fault-tolerant machines are expected to need physical qubits numbering in the hundreds of thousands, possibly millions. Every result above still counts physical qubits in the tens or low hundreds. Noise and decoherence, the process by which a qubit loses its quantum state through heat, electromagnetic interference, or even stray cosmic rays, get harder to manage as chips grow, not easier, which is exactly why below threshold behavior mattered so much on Willow. Verifying that a quantum algorithm’s output is actually correct, especially on problems too large to check classically, is its own unsolved research area. Then there’s post-quantum cryptography.
In August 2024, NIST finalized three post-quantum encryption standards, FIPS 203 (ML-KEM), FIPS 204 (ML-DSA), and FIPS 205 (SLH-DSA), closing out an eight-year evaluation process, and is urging organizations to begin the transition now rather than waiting for a quantum computer capable of breaking RSA and elliptic-curve encryption through Shor’s algorithm. Finally, the field faces a real talent shortage and a high cost of entry, since building and running dilution refrigerators, ion traps, or photonic systems takes specialized engineering that few universities or companies can staff at scale.
Progress vs Challenges: The Verdict
So did 2024 bring quantum computing meaningfully closer to fault tolerance? Yes, on the specific and narrow question of error correction. Willow proved below threshold scaling works in practice, not just in theory. Microsoft and Quantinuum showed logical qubits can already beat physical qubits by a wide margin on a small system. The Harvard, MIT, and QuEra team showed dozens of logical qubits can be integrated at once. These are genuine, peer-reviewed, independently verified results, not marketing claims.
None of them add up to a computer that outperforms classical machines on a problem anyone actually cares about, though. Qubit counts are still small. Every real-world pilot, in medicine, chemistry, or finance, is still a proof of concept running alongside classical benchmarks, not replacing them. The honest read on quantum computing 2024 is that the error correction problem, long considered the field’s central obstacle, took a real hit. The scaling problem did not.
What’s Next: 2025 and 2026 (Updated September 2026)
The pace hasn’t slowed since 2024, and several of the open threads above have already moved forward.
Topological qubits got a second, very different demonstration. In February 2025, Microsoft unveiled Majorana 1, an eight-qubit chip built on a new class of material it calls a topoconductor, designed around Majorana zero modes rather than the anyon-based approach Quantinuum had shown in late 2024. Microsoft framed it as a first step on a roadmap toward a million qubits on a single chip, and by later in 2025 the team reported having performed the two parity measurements needed to actually operate a Majorana-based qubit. The claim remains genuinely contested, though: Microsoft’s own Majorana research had been walked back once before, in a retracted 2021 paper, and outside groups, including a University of New South Wales team, published work in 2025 questioning whether current decoherence times are long enough to support the approach. Worth watching, not yet settled.
Google crossed from “below threshold” to “verifiable quantum advantage.” In October 2025, Google Quantum AI reported that Willow ran a new algorithm called Quantum Echoes, which uses out-of-time-order correlators to probe how information scrambles through a quantum system, about 13,000 times faster than the best classical estimate on a comparable supercomputer, and — critically — in a way that can be independently reproduced and checked on other hardware, unlike the harder-to-verify 2019 sampling claim. Google called it the first verifiable quantum advantage in the field’s history. Some researchers quoted in Nature’s own coverage cautioned that it isn’t fully settled whether a smarter classical algorithm could eventually close the gap. It’s also still a narrow, physics-relevant benchmark, not a commercially useful computation.
Google added a second hardware bet. In March 2026, Google Quantum AI announced it was expanding beyond its decade-long focus on superconducting qubits to build a dedicated neutral atom hardware program, hiring physicist Adam Kaufman to lead it. The company framed this as a way to pursue both approaches’ strengths in parallel: superconducting qubits run very deep circuits quickly, while neutral atom arrays currently scale to larger qubit counts more easily. Google says it’s still targeting a commercially useful, error-corrected machine by the end of the decade.
Topological error protection reached a further milestone. In July 2026, a team from Harvard, Quantinuum, Stony Brook University, and the University of Chicago published the first demonstration of a universal topological gate set, built by combining anyon braiding with a second operation called fusion on a 54-qubit version of Quantinuum’s H2 processor. The result matters because braiding alone hadn’t been enough to reach universality on the simplest non-Abelian systems; adding fusion as a computational primitive closed that gap and let the team prepare a “magic state” directly through topological operations, without the resource-heavy distillation process most error-corrected designs currently rely on.
The broader field kept diversifying through 2026, with steady incremental progress rather than another single headline result: new qubit fabrication techniques aimed at making superconducting chips easier to manufacture at scale, continued hybrid quantum-classical chemistry work from companies like IonQ, and growing interest in early-stage approaches like magnon-based and photonic qubits that trade current performance for a possible path around today’s scaling and cooling bottlenecks.
The overall direction is consistent with what 2024 suggested: more architectures tested in parallel, larger logical qubit and gate-set demonstrations, and a slow, uneven move from single proof-of-concept results toward repeatable, verifiable ones. Anyone evaluating this space now needs to track hardware announcements alongside error correction and verification papers, since the two no longer move separately, and needs to hold Google’s and Microsoft’s more ambitious claims against the same scrutiny this article applied to the 2024 results: real progress, real limits, both true at once.
FAQs
Is quantum computing useful today? For a small number of narrow research problems, yes, in a limited pilot sense. Chemistry, materials science, and mRNA structure prediction have all seen genuine quantum-assisted results. For general business computing, no. Classical computers still outperform quantum machines on nearly every practical task.
Is today’s encrypted data at risk from quantum computers? Not from a working quantum computer today, since no machine can currently break RSA or elliptic-curve encryption. The real risk is “harvest now, decrypt later,” where encrypted data gets collected today and decrypted once a large enough quantum computer exists years from now. This is exactly why NIST finalized its post-quantum cryptography standards in August 2024.
What was the biggest breakthrough in quantum computing 2024? Most researchers point to error correction crossing the below threshold barrier, achieved independently by Google’s Willow chip and the Harvard, MIT, and QuEra neutral atom team, both recognized jointly by Physics World as its 2024 Breakthrough of the Year.
What’s the biggest obstacle left? Scaling. Current systems run on qubit counts in the tens or low hundreds. Practical fault-tolerant computing is expected to need physical qubits numbering in the hundreds of thousands or more. Google’s October 2025 verifiable quantum advantage result and the July 2026 topological gate-set demonstration are real steps, but neither changes that underlying qubit-count math.
Which industry will benefit first? Chemistry, materials science, and pharmaceuticals are furthest along, based on pilots like the IBM and Moderna mRNA work and Microsoft’s battery material research with PNNL. Finance and cybersecurity are close behind, mainly through post-quantum cryptography and quantum-secured networking rather than quantum computation itself.
Conclusion
The latest breakthroughs in quantum computing 2024 are real. Below threshold error correction, an 800 times more reliable logical qubit, a 48 logical qubit neutral atom processor, and a first working topological qubit are not incremental updates. They’re results that took the field close to three decades to reach. At the same time, none of them turn a quantum computer into a tool that beats classical computing on a problem that matters commercially, not yet. The honest position on quantum computing 2024 sits between the two extremes you’ll find in most headlines. Real progress happened. Real limits remain. Both statements are true at once, and anyone making decisions based on this field should hold onto that balance rather than picking a side — a balance that has held up well through everything the field has done since.
Senior SEO Content Marketing Manager at Trendusai.com
Rashida Hanif is a Senior SEO Content Marketing Manager, specializing in data-driven content strategy and SEO. She helps brands improve online visibility through keyword research, content planning, and AI-powered marketing insights.
