47 stories
Show: Summaries / Headlines only
Ranked by logical qubits, fault-tolerant applications look like a staircase. Ranked by the gates each one needs, the list within reach of announced machines is shorter. Elliptic-curve attacks come before every molecular ground-state estimate in Google's 2025 tables.
Are the qubits physical or logical? Is there a paper, and who checked it? A reader can judge a quantum computing announcement with seven questions, each shown here with a 2026 case and the document that answers it.
Alice & Bob's new count is about a quarter of its 2024 estimate. The authors assumed cat qubits that keep their value for about 100 years. Researchers have measured up to 22 seconds under full control.
The Sydney company said it cut its February estimate by about an order of magnitude in qubit-time with its Pinnacle 1.2 update. It has not published the circuits, the decoder, or the hardware behind its decoding speed.
Quantinuum has reported logical-qubit counts of 94, 50, 48, and one on the same machine with 98 physical qubits, from four different experiments. With five questions, a reader can tell what most such announcements demonstrated.
Quantinuum's team cut logical errors about 12-fold with each step up in code distance on the 98-qubit trapped-ion machine, for one stored logical qubit. The preprint has not been peer reviewed.
An open-weight model running on IBM's GPUs found a 52-fold faster design for about $26 in inference. The speedups are synthesis-tool estimates for an FPGA decoder, from a single run of each configuration.
Google's team measured the error of an encoded T state at 1 to 2 in 10,000 or less. Gidney's 2025 RSA-2048 estimate assumes cultivated states about 1,000 times cleaner, followed by a further distillation step.
DOE drew the problems from chemistry, materials, subatomic physics, and applied mathematics. It wrote that it did not set target problem sizes and chose problems that first-generation machines could tackle in small demonstrations.
The team ran circuits with nearly 5,000 two-qubit gates and 5,000 mid-circuit resets and reproduced the predicted physics without error mitigation. It encoded no logical qubits, so it did not test error correction.
