Two Answers to Quantum's Wiring Problem, and One Headline That Had to Be Corrected

by Karl Lindström · September 1, 2026

Semiconductors & Hardware Reporter, Frontier Tech Wire

Two announcements landed on Monday about the least photogenic part of a quantum computer: the cables that go into the cold, and the box that holds the cold. Neither involves a qubit count worth boasting about. Both are about whether any of this fits in a rack.

QTREX Quantum Ltd., which trades on Nasdaq under QTEX and is based in Ness Ziona, said at 9:07 a.m. ET on August 31 that it will unveil an interconnect architecture supporting 17,280 coaxial lines per cryogenic stage in full-scale commercial systems. The company said it will show a physical demonstrator at IEEE Quantum Week 2026, running September 13 to 18 at the Metro Toronto Convention Centre, from booth 712, with the presentation set for September 13.

The architectural idea is straightforward to state. Rather than running a conventional bundle of discrete coaxial cables, connectors and thermal anchoring hardware down through a dilution refrigerator, QTREX distributes lines around the circumference and across the surface of each cryogenic stage, using the plate area as well as the perimeter. The interconnect elements themselves are built with what the company calls Additively Manufactured Electronics, a process it says integrates conductors, dielectrics, shielding and mechanical routing into monolithic structures.

The comparative claim is where the release becomes interesting. Its exact wording reads: "Across the publicly available product specifications, technical publications and industry roadmaps reviewed by QTREX, no other disclosed physical interconnect architecture for dilution refrigerators, regardless of manufacturing technology, reaches even half of this capacity at each cryogenic stage." That is a claim about a literature review QTREX conducted and did not publish. No competing system is named, no specification list is appended, and no third party is cited as having verified the comparison.

Thirty-six minutes later, at 9:43 a.m. ET, GlobeNewswire carried a corrected version of the announcement under a materially different headline: "Correction: QTREX Establishes Cryogenic Interconnect Capacity, More Than Four Times the Highest Publicly Disclosed Industry Benchmark." The corrected release does not state what was corrected. Its body text still carries, word for word, the same sentence about no other architecture reaching even half the capacity, and the phrase "four times" appears nowhere in that body. So the headline of the corrected release asserts a multiple of more than four, while the body of the same release supports a multiple of more than two.

None of that is evidence of bad faith. Issuing a correction through the wire is the ordinary mechanism for fixing a release after distribution, companies use it routinely for reasons as mundane as a mistyped date, and QTREX may well hold review data that reconciles the two figures. The point is narrower and entirely about what a reader can check: the two numbers are not reconciled in either document, both rest on the same unpublished internal review, and the release does not say which one supersedes the other. A reader taking the headline at face value would be relying on a multiple that the accompanying text does not support.

What neither version contains is the engineering data that would let a customer evaluate the architecture. There is no heat load per line, no figure for thermal budget at any stage, no attenuation or crosstalk data, no stated temperature for the stages in question, and no measurement conditions of any kind. There is no peer-reviewed paper and no preprint attached. The Toronto exhibit is described in the release as a scaled technology demonstrator that reproduces the architecture's density and demonstrates its implementation across multiple temperature stages of a dilution refrigerator, which is a statement about geometry rather than about signal integrity or cooling performance at millikelvin, and a scaled one at that rather than the full 17,280-line configuration.

Chief executive Dagi Ben Noon put the pitch in fault-tolerance terms. "By unveiling an architecture capable of supporting more than 17,000 coaxial lines at each cryogenic stage, we are providing the scalable physical interconnect infrastructure required for fault-tolerant quantum computing," he said in the release. "In Toronto, we will turn this capability into a physical demonstrator that industry participants can inspect and configure around their needs." The company said it is already engaged with quantum computing companies, US federal laboratories, academic research institutions and defence organisations, but named no individual customer or partner among them.

Two details from the boilerplate are worth carrying into any read of the story. QTREX is a foreign private issuer whose forward-looking statements point readers to its Form 20-F risk factors, and its described business is not purely quantum: alongside the cryogenic connectivity platform, the company lists a medical technology portfolio spanning respiratory support and blood monitoring systems. Investors sizing the quantum opportunity here are sizing part of a company, not all of it.

The other Monday announcement approaches the same bottleneck from the opposite end. Diraq and Equinix said in a release datelined Sydney on August 31 that they will install a silicon spin quantum computer inside Equinix's Sydney data centre, with installation targeted for October 2026. The machine is an eight-qubit silicon chip drawing less than 20 kilowatts, with self-contained cryogenic cooling and control electronics. Financial terms were not disclosed, and neither company put a value on the arrangement.

Andrew Dzurak, Diraq's founder and chief executive, was direct about the framing. "Quantum computers are about to become as essential to data centers and computing infrastructure as data servers, CPUs and GPUs," he said, adding that "the data center is where quantum computing goes mainstream, and that shift starts now." Jarrod Nink, Equinix's managing director for Australia, said the collaboration will demonstrate how quantum computing can be securely deployed alongside AI, cloud and high-performance computing environments.

Diraq describes itself as commercialising quantum computing using existing CMOS processes and commercial semiconductor foundries, with quantum dot technology derived from roughly two decades of Dzurak's research and an eventual goal of millions of qubits on a single chip. The company was founded in Sydney, is headquartered in Palo Alto, and says it has more than 100 team members globally, with operations in Melbourne, Palo Alto, Los Angeles and Chicago. Neither release states its ownership status, and no investors are named.

The claims that go unquantified here are the same ones that matter most. Diraq published no qubit fidelity, no gate error rate and no coherence time for the eight-qubit chip going into the Equinix facility, and its statement that higher qubit counts will require only a chip replacement with no infrastructure changes is a company assertion about a machine that has not yet been installed. Under 20 kilowatts is a genuinely useful number because it is comparable to a dense conventional rack, but power draw says nothing about whether the qubits work.

Put the two releases side by side and the shape of the sector's next problem is visible. One company is building infrastructure for more than seventeen thousand control lines per cryogenic stage; the other is shipping eight qubits into a live commercial building. Both are correct about where the industry is heading, and the distance between them is the reason cryogenic plumbing has quietly become an investable category. Nothing here is investment advice.