British quantum computing company Moth announced on Monday, November 16, 2026, Tessa, a method that loads classical data onto an entire quantum processor at once — and reads it back. In tests on the 120-qubit IBM Quantum Nighthawk r2 processor, Tessa encoded and recovered 1,568 values, more than 12 times the previous record for this kind of experiment, which stood at 128 values.
Read also: Yonsei switches on IBM Nighthawk quantum computer to speed up drug discovery
Loading data is one of quantum computing's biggest bottlenecks: before any calculation, information such as images, audio or DNA sequences must be written into the qubits, and every step of that process adds noise. Earlier methods also assumed each data qubit was wired directly to every addressing qubit — something that does not happen on IBM's superconducting chips, where each qubit connects only to its nearest neighbours.
How Tessa works
Tessa works around the problem by using the chip's wiring as it is. Qubits are split into two groups, like a chessboard: data qubits, which hold the information, and neighbouring address qubits, which tell each data qubit where every value belongs. That removes the need for SWAP gates to move information around the chip or for helper (ancilla) qubits — and all data qubits are written at the same time, so circuit depth does not grow with chip size: encoding 784 values on the Nighthawk r2 took only 19 layers of two-qubit gates.
What was tested
In the experiments, the team encoded a pixel-art cat (the “QCAT”), with a correlation of 0.95 or better between the original image and the one read back, as well as an audio waveform, musical scores and biological sequences — a 784-letter stretch of mitochondrial DNA came back with 92% of its letters correct, and protein sequences, including p53, were also tested. On a 156-qubit Heron r3 processor, Tessa encoded and decoded 441 numbers. Storing a second value in the phase of each qubit raised the Nighthawk total to 1,568 values.
The method's limits
There is a trade-off: the more address neighbours a data qubit has, the more repeated measurements are needed to recover the values at the same accuracy — and the 1,568-value run was less accurate (correlation of 0.69) than the 784-value run. Even so, the result tackles the main practical obstacle for quantum machine learning, the cost of loading classical data onto today's hardware. The scientific paper, Encoding Classical Data Across an Entire Quantum Processor, by Daniel Bultrini, is on arXiv, and Tessa is already available to try on Moth's Atlas platform.
Imagem: four pixel-art images before and after being encoded on IBM's Nighthawk r2 quantum processor with the Tessa method. Image: Moth; artwork by M. Fish and M.A. Kozhikkal (SalvaPixel).
Fontes
- Tessa: loading a record amount of data onto real quantum chips — Moth (official source), November 16, 2026
- A Breakthrough That Really Counts — Moth release via PR Newswire, November 16, 2026
- From Cat Pictures to DNA, Moth Quantum Reports Progress on Getting Everyday Data Into Quantum Computers — The Quantum Insider, November 16, 2026
- Moth's Tessa Encodes and Recovers 1,568 Values on a 120-Qubit Chip — Unite.AI, November 16, 2026
Transparency: this article was produced with the assistance of artificial intelligence, based on cross-checking public posts and the internet sources cited above. We recommend consulting the original sources to confirm the details.
By Muse