Quantum computing headlines arrive almost every month, each claiming a "breakthrough." But what is a quantum computer actually doing differently, and which recent milestones genuinely matter? This plain-English guide explains the basics in a few minutes and then walks through the most important verified achievements from late 2024 to 2025, plus what the big players say comes next. Facts are current as of October 2026.
Quantum Computing in Plain English
A normal computer stores information in bits, which are either 0 or 1. A quantum computer uses qubits. Thanks to superposition, a qubit can be in a blend of 0 and 1 at the same time, and thanks to entanglement, the states of many qubits can be linked so that they behave as one system. A well-designed quantum algorithm uses interference to amplify the right answers and cancel out the wrong ones.
That does not make quantum computers faster at everything. They are expected to shine at specific problems: simulating molecules and materials, certain optimisation tasks and some areas of mathematics, including the factoring problems behind parts of today's encryption. For browsing the web or editing a spreadsheet, a classical computer will always be the better tool.
The catch is fragility. Qubits lose their quantum state when disturbed by heat, vibration or stray signals, which introduces errors. Fixing that, through quantum error correction, is the central challenge of the field.
The Big Problem: Errors, and Why "Logical Qubits" Matter
Error correction spreads one piece of information across many physical qubits to form a single, more reliable "logical qubit." In theory, adding more physical qubits should make the logical qubit better. In practice, for years, adding qubits also added more noise, so things got worse. A key goal was to show error rates dropping as the system grows, a regime researchers call operating "below threshold."
Recent Milestones That Actually Matter
December 2024: Google's Willow chip. Google Quantum AI announced Willow, a 105-qubit superconducting chip. Its headline result was error correction that improved with scale: as Google moved from 3x3 to 5x5 to 7x7 grids of qubits, the error rate was cut roughly in half at each step, demonstrating below-threshold behaviour. Willow also completed a random circuit sampling benchmark in under five minutes, a task Google estimates would take one of the fastest supercomputers around 10 septillion years. That benchmark has no practical use, but it shows raw capability.
June 2025: IBM's fault-tolerance roadmap. IBM laid out a plan to build Starling, a large-scale, fault-tolerant quantum computer, at a new quantum data center in Poughkeepsie, New York, targeting 2029. IBM says Starling is designed to run 100 million quantum operations using 200 logical qubits, with a successor called Blue Jay aimed at 2,000 logical qubits. Its approach relies on qLDPC error-correcting codes, which IBM estimates cut error-correction overhead by roughly 90%. Intermediate processors are named Loon (2025), Kookaburra (2026) and Cockatoo (2027). These are targets, not finished machines.
October 2025: a verifiable quantum advantage. Google reported in Nature an algorithm called Quantum Echoes that ran about 13,000 times faster on Willow than the best known classical method on a top supercomputer. Unlike earlier benchmarks, the result can be checked by repeating it on another quantum computer of similar quality. In a proof-of-principle experiment with UC Berkeley, the technique was used to study molecules with 15 and 28 atoms, with results matching traditional NMR measurements.
October 2025: a Nobel Prize for the foundations. The 2025 Nobel Prize in Physics went to John Clarke, Michel H. Devoret and John M. Martinis "for the discovery of macroscopic quantum mechanical tunnelling and energy quantisation in an electric circuit." Their 1980s experiments underpin today's superconducting qubits.
What Comes Next, and How to Read the Hype
The next big goal is a machine that solves a useful real-world problem faster than any classical computer while keeping errors under control for long calculations. Companies have roadmaps pointing to the end of this decade, but roadmaps can slip. A few tips for reading quantum news:
- Logical beats physical. A big physical qubit count means little if the error rate is high. Look for logical qubit numbers and error rates.
- Ask "useful or benchmark?" Many speed records use tasks designed to be hard for classical computers, not tasks anyone needs solved.
- Check verification. Can the result be confirmed independently? Peer-reviewed papers carry more weight than press releases.
- Separate plans from results. "Will build by 2029" is a plan, not an achievement.
Conclusion
Quantum computing has moved from physics lab curiosity to serious engineering. Recent milestones show error correction finally working in the right direction and the first checkable advantages on scientific problems. Practical, fault-tolerant machines are still ahead, but the path toward them is clearer than ever. Keep an eye on logical qubits and verified results, and you will be able to tell real progress from noise.
Sources
- Google: Meet Willow, our state-of-the-art quantum chip
- UC Berkeley (reposted from Google): Our Quantum Echoes algorithm is a big step toward real-world applications
- IBM Newsroom: IBM Sets the Course to Build World's First Large-Scale, Fault-Tolerant Quantum Computer
- NobelPrize.org: The Nobel Prize in Physics 2025