TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
PsiQuantum has demonstrated an 8×8 optical circuit switch built on its silicon photonics platform, reporting 1.2 dB insertion loss, 40 dB crosstalk performance and reconfiguration in under 3 milliseconds. The company is sharing samples with selected prospective customers, but has not yet announced a commercial product or independent test results.
PsiQuantum has demonstrated an 8×8 optical circuit switch for potential use in AI data center networks, reporting lab measurements of 1.2 dB insertion loss, 40 dB crosstalk performance and less than 3 milliseconds to reconfigure. The quantum-computing company says it is sharing demonstration samples with selected prospective customers as it develops commercial products, though it has not announced a release date or customer deployments.
The demonstration vehicle is built on PsiQuantum’s low-loss silicon photonics platform, which the company says it developed over nearly a decade for utility-scale quantum computing. PsiQuantum reported that testing in its own laboratories produced typical insertion loss of 1.2 dB, crosstalk performance of 40 dB and switching reconfiguration times below 3 milliseconds. The company said the results exceeded its initial targets and were achieved without optical amplification.
The device’s 8×8 radix describes its switching configuration. PsiQuantum has not published a full technical data sheet, test protocol or independent validation in the material provided. The figures are therefore company-reported laboratory results, rather than evidence of performance in a deployed data center. PsiQuantum says prospective customers are receiving samples of the demonstration vehicle while work continues toward products it expects to announce in the coming months.
The company says the initial optical switching products do not require barium titanate (BTO), a material it has also integrated at 300-millimeter wafer scale. PsiQuantum says it uses a 64×64 BTO switch in its quantum systems, but did not say that this switch is part of the AI data center demonstration. The announcement presents the silicon photonics platform as the basis for the current device, with BTO described as a possible route to more advanced future products.
Optical Links for Expanding AI Networks
AI data centers connect large groups of computing accelerators, and moving data between them can place increasing demands on network capacity and efficiency. Optical circuit switches direct light between network paths, offering a way to reconfigure connections without relying solely on conventional electronic switching. PsiQuantum is positioning its device as one potential component in that infrastructure, not as a complete data center networking system.
If the reported performance carries over to commercial hardware and practical deployments, the switch could give operators another option for connecting compute resources as AI workloads grow. The announcement does not establish savings in power, cost or latency for customers, nor does it show that the device has been integrated into an operating AI data center. Those questions will depend on product specifications, compatibility, reliability and customer testing.
The development also offers PsiQuantum a potential application for technology created for quantum computing while it continues that longer-term effort. CEO Victor Peng said the opportunity arose from work on “the hardest scaling problems” in the company’s quantum roadmap, and described the AI infrastructure opportunity as a way to put its technology to work now. That is the company’s rationale; the announcement gives no revenue forecasts or commercial commitments.
Products Worth Considering
Bose S1 Pro+ Portable PA Speaker
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
Jackery Solar Generator 1000 v2 with 200W Solar Panel
As an affiliate, we earn on qualifying purchases.
A Quantum Platform Finds Another Use
PsiQuantum’s core objective is to build fault-tolerant quantum computers, systems that would require large numbers of components and precise control. The company says the requirements it pursued for that work exceed many of those needed for optical switching in AI data centers. It is using the announcement to argue that its silicon photonics work can serve both fields, while continuing its quantum-computing program.
The announcement follows a September 2026 development: PsiQuantum said it had signed definitive documentation with the U.S. Department of Commerce to accelerate research and development, domestic manufacturability and performance of critical components, including optical switches. The source material does not specify whether that agreement funds or directly supports this particular 8×8 demonstration.
Rob Soderbery, PsiQuantum’s executive vice president, said the growth and complexity of AI computing infrastructure will increase the role of optics. His statement reflects the company’s view of the market opportunity. The announcement does not identify a customer, data center operator or network equipment partner involved in the demonstration.
“This opportunity exists because we chose to solve the hardest scaling problems at the beginning of our roadmap to utility-scale quantum computing.”
— Victor Peng, PsiQuantum chief executive
Deployment Evidence Is Still Missing
The reported measurements come from PsiQuantum laboratory testing. The company has not supplied independent test results, detailed measurement conditions or information on how the demonstration performs under sustained data center workloads. It is also unclear how the device compares with alternatives on throughput, power consumption, reliability, cost and ease of integration.
PsiQuantum has not named prospective customers receiving samples, stated how many samples are being shared or announced a commercial launch date. The company says product announcements are expected in the coming months, but has not provided pricing, production volumes or deployment plans. The relationship between the September Department of Commerce documentation and this specific product effort is also not detailed in the source material.
Customer Samples and Product Plans
The next stated step is evaluation of demonstration samples by selected prospective customers. PsiQuantum says it is advancing the platform toward commercial products and expects to announce those products in the coming months. Any follow-up announcement may clarify product specifications, availability and intended network configurations.
For now, the key milestones to watch are customer testing, independent or deployment-level performance evidence, and details on manufacturing and commercial readiness. PsiQuantum has not confirmed when any customer installation will take place.
Key Questions
What did PsiQuantum announce?
PsiQuantum disclosed an 8×8 optical circuit switching demonstration based on its silicon photonics platform and said it is developing commercial products for AI data center connectivity.
What performance figures did the company report?
PsiQuantum reported typical laboratory results of 1.2 dB insertion loss, 40 dB crosstalk performance and reconfiguration in less than 3 milliseconds, without optical amplification. These are company-reported results, not independently validated figures in the available source material.
Is the switch available to data center operators now?
No commercial availability was announced. PsiQuantum says it is sharing demonstration samples with selected prospective customers and expects to announce commercial products in the coming months.
Does the demonstration use barium titanate?
PsiQuantum says BTO is not required for the demonstration vehicle or its initial products. The company describes BTO integration as an additional path for faster, more advanced future products.
Has the switch been tested in a working AI data center?
The announcement describes testing conducted in PsiQuantum’s laboratories and sample sharing with prospective customers. It does not report a data center deployment or customer test results.
Source: rss
