Error correction eats most of the qubits in a future quantum computer. Photonic Inc. says it has a way to cut that cost. The company reports that its SHYPS code family can run logic with fewer physical qubits than the standard surface code, at the code sizes it tested. It published a peer-reviewed paper in Nature Communications, announced August 25.
“Efficient QLDPC logic is no longer a theoretical promise,” said Dr. Stephanie Simmons, Photonic’s chief quantum officer, in the company’s press release. “It’s a demonstrated result, with real implications for architectures and timelines.”
What’s new
- A peer-reviewed Nature Communications paper on SHYPS QLDPC codes, announced August 25, 2026.
- Photonic says SHYPS can run logic with meaningfully fewer physical qubits than the surface code, at the code sizes tested.
- Photonic calls it the first QLDPC code family to show efficient logic, not just storage. That priority claim is the company’s own.
What stays the same or isn’t confirmed
- No qubit counts or ratios in the release, only “meaningfully fewer.”
- No stated commercial date, roadmap or deployment schedule.
- The release doesn’t say whether results came from hardware or simulation. The preprint we read describes simulation only.
- Larger codes still need to be shown.
The overhead problem
Quantum computers need many physical qubits to protect one reliable “logical” qubit from errors. That overhead is a major reason fault-tolerant machines are still years away. Surface codes power most of today’s error-corrected quantum chips, but they waste qubits: each logical qubit is spread across a large grid of physical ones just to catch and fix errors. Researchers have studied quantum low-density parity-check codes for years, and theory says they should need fewer physical qubits per logical qubit than practical surface codes. The open problem was running logic on them efficiently. Photonic says SHYPS addresses it, and the company’s announcement frames it that way.
How SHYPS works
SHYPS stands for Subsystem Hypergraph Product Simplex. The codes handle error correction and computation together, in the same structure, instead of treating them as separate jobs. That lets Photonic run logic gates directly on protected qubits, rather than shuttling data between a storage code and a separate computing step.
The preprint version of the paper describes circuit-level simulations, not a hardware run. The authors report near-memory performance for logic circuits 126 layers deep. The release says the gains need high-connectivity hardware and points to Photonic’s Entanglement First architecture, which links silicon spin qubits optically. The sources we read don’t describe a hardware test of SHYPS. We did not read the final journal paper.
The release says SHYPS is competitive with surface codes on logical clock time and performance, which is parity, not a clear win. The preprint’s main comparisons are asymptotic, and the authors note that constant factors can matter. Some of its figures are estimates for one example, not measurements, as in the table below. The coverage we read, from The Quantum Insider, leans on the company’s press release, and we found no independent expert assessment.
| Logical gate | Syndrome rounds, SHYPS | Comparison | Source |
|---|---|---|---|
| CNOT | 4 | About 40 with lattice surgery (authors’ estimate, distance 20) | arXiv preprint |
| CZ | 4 | Not given | arXiv preprint |
| S | 6 | Not given | arXiv preprint |
| H | 8 | Not given | arXiv preprint |
Why It Matters to You
- Every physical qubit saved on error correction is one a company can spend on computation instead, and one it doesn’t have to build, cool and control. A fault-tolerant machine could end up smaller and simpler. IBM has described modular cryogenic fridges as another answer to the same scaling pressure.
- For a small university lab (picture Elena, an illustrative example, not a real person), cheaper error correction might someday lower the cost of running protected computations. Photonic gives no qubit-count ratio, price or access plan, so that is a possibility, not a forecast.
- Nothing here is available to buy or use today. Researchers and quantum hardware builders would feel any benefit first.
Decoders attack the same overhead problem from another angle. Our coverage reported that an AI decoder trained on synthetic data topped matching benchmarks, the company says, and that IonQ reported its decoder kept pace on a laptop chip, in simulation.
Quick answers
What does SHYPS stand for? Subsystem Hypergraph Product Simplex. It’s a family of quantum low-density parity-check (QLDPC) codes from Photonic Inc.
How many qubits does it save? Photonic hasn’t published a figure in its release. It says SHYPS needs meaningfully fewer physical qubits than surface codes at the code sizes tested, but the exact savings aren’t given.
Was it tested on a real quantum computer? The preprint we read describes simulations only, and the press release doesn’t say. Check the published paper for the final answer.
When could this reach commercial machines? Photonic has given no date. The release says only that the results have “real implications for architectures and timelines.”
Sources
- GlobeNewswire: Photonic’s Breakthrough Demonstration of Efficient QLDPC Logic Published in Nature Communications (Aug 25, 2026)
- arXiv: Computing Efficiently in QLDPC Codes (preprint, 2025)
- The Quantum Insider: Photonic Publishes SHYPS QLDPC Code Results in Nature Communications (Aug 26, 2026)
