QTREX says its cryogenic interconnect beat market leading cables by 60 dB in independent tests

Started by Badger, Oct 03, 2026, 10:25 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Topic: QTREX says its cryogenic interconnect beat market leading cables by 60 dB in independent tests   Views(Read 105 times)

Badger

Israeli company QTREX Quantum has announced test results for a new cryogenic interconnect designed for quantum computers. The product is made with additive manufacturing, essentially 3D printing electronics, and packs fully shielded channels about 0.34mm apart. That works out at around 30 channels per centimetre of width. Wiring is one of the biggest headaches in scaling superconducting quantum computers, so this is an interesting area. Superconducting machines need several lines per qubit, so the numbers add up quickly

The tests were carried out by an unnamed leading quantum computing company using its own fridge, instruments, reference cables and calibration at 20 millikelvin, the temperature superconducting processors run at. According to QTREX, the channel isolation was better than 100 dB across the 4 to 8 GHz band between neighbouring channels. That is 60 dB better than the market leading flexible cryogenic cabling. Channels also stayed phase matched at that temperature

The other figures look solid too. Insertion loss was about 1 dB at 1 GHz and 2 dB at 6 GHz including connectors, and the interconnect conducts roughly half the heat of equivalent coaxial cable. Heat matters enormously, because every cable brings warmth down into the fridge. Performance reportedly stayed the same through repeated cooldowns, bending and vacuum exposure

Chief executive Dagi Ben-Noon said 60 decibels more isolation than leading flexible cabling represents not an improvement but a different category. The company says it is finalising a binding agreement with the testing company and expects to name it within weeks. It also expects more quantum computing partnerships in the final quarter of the year

This links back to the Q-Factor piece and the 3D model of a superconducting computer, where the amount of wiring was obvious. Hand assembled coaxial cables will not get anyone to a million qubits. I would like to see the tester named before getting too excited, but the numbers are impressive if they hold up. Is wiring the most underrated problem in quantum computing?. Companies like this tend to live or die on their first big customer

Powerbombed my keyboard, it deserved it

Taker04

Sixty dB is a factor of a million in power terms. If that holds up, crosstalk between channels almost stops being an issue

Crosstalk is one of the sources of errors people forget about. Cleaner signals mean better gate fidelity
It's not a bug, it's a feature

NightCrawler

Surviving repeated cooldowns and bending is a practical detail that matters. Cables that crack after a few thermal cycles are a nightmare to replace inside a dilution fridge. Each warm up and cool down takes days

Arkham93

Unnamed leading company is the part that makes me cautious. Press releases with anonymous partners are common in this industry. Let us see who it is first. QTREX says the name is coming within weeks, so we will not wait long

Messi

Wiring is definitely the underrated problem. Every photo of a quantum computer shows a forest of cables. You cannot fit a million qubits' worth of coax into a fridge. Multiplexing signals is the other route people are exploring

Hannah

Additive manufacturing for electronics is an interesting approach. Printing dense interconnects avoids the hand assembly that makes coax so expensive

It could bring costs down a lot if it scales. Prototyping new layouts would also be quicker. Iteration speed matters a lot in hardware

StarfleetCaptain

Half the heat load is just as important as the isolation. Fridges have limited cooling power at the coldest stage

Less heat per channel means more channels per fridge. Cooling power at the mixing chamber is measured in microwatts

Related Topics (1)