Short answer: Gottesman-Kitaev-Preskill (GKP) states encode a qubit in light in a way that supports error correction, but they are hard to produce. A 2025 Nature paper reports generating GKP qubit states on an integrated silicon nitride photonic chip, with at least four resolvable peaks in both quadratures and a 3 by 3 grid of negative Wigner function regions. A 2024 Physical Review A paper proposes a scheme with success probabilities two orders of magnitude higher than purely photon-counting methods. The experiment is a source demonstration, and the authors say further loss reduction is needed.
Why GKP states matter
The Nature abstract describes GKP states as one of the most attractive qubit encodings for a photonic quantum computer, because they allow universal gate sets using deterministic, room-temperature-compatible Gaussian operations. Earlier demonstrations of optical GKP and other non-Gaussian states relied on free-space optical components, which the authors say hinders the scaling needed for a utility-scale system.
The 2025 integrated-chip experiment
The authors use an ultra-low-loss integrated photonic chip made on a customized multilayer silicon nitride 300-millimeter wafer platform, connected by fibre to high-efficiency photon-number-resolving detectors. They report generating GKP qubit states that show mode-level features needed for fault tolerance: at least four resolvable peaks in both the p and q quadratures, and a clear lattice of negative Wigner function regions, in this case a 3 by 3 grid.
They add that the states show enough structure to indicate that the devices could, after further reduction in optical losses, yield states for the fault-tolerant regime. That is the authors' assessment, and it means the generated states are not yet at that level. They describe the experiment as validating a key pillar of bosonic architectures and paving the way for arrays of GKP sources.
A proposed higher-probability method
The Physical Review A abstract (2024) says measurement-based methods, in which part of a set of entangled squeezed vacuum modes is measured with photon-number-resolving detectors, are leading candidates but suffer from low success probabilities. Photon-number-resolving measurements both supply the nonlinearity needed for quality states and drive down the probability of success.
The authors propose breeding approximate squeezed coherent state superpositions created by generalized photon subtraction. This supplements two photon-number-resolving measurements with a single high-success-probability homodyne measurement. They report success probabilities two orders of magnitude higher than strictly photon-number-resolving methods, with states whose error-correction capability is equivalent to up to a 10 dB squeezed GKP qubit. This is a theoretical proposal.
How the two relate
The experiment and the proposal address different parts of the same problem. The chip work focuses on producing states with low loss at a scalable fabrication scale, while the proposal focuses on raising the chance that a heralded state is produced at all. Neither abstract claims to combine the two.
Limits
The chip experiment generates states and does not report a logical gate or error-corrected computation. The Physical Review A result is a calculation. The exact success probability values did not render in the retrieved text, so none is quoted here.
FAQ
What is a GKP state?
A bosonic encoding of a qubit in light with built-in error-correction ability, per the 2025 abstract.
What did the 2025 experiment show?
GKP states on a silicon nitride chip with a 3 by 3 grid of negative Wigner regions.
Is it fault tolerant yet?
No. The authors say further loss reduction is needed.
How does the 2024 proposal differ?
It adds a homodyne measurement to raise the success probability.
Is there a photonic quantum computer built this way?
Not according to these abstracts.
All field notes