DiCon launches 300×300 and 64×64 optical circuit switches for AI data centers

DiCon Fiberoptics has expanded its MEMS optical circuit switch (OCS) lineup with two new, value-focused models: a 300×300 and a 64×64, aimed at AI data centers using leaf-and-spine networks and AI server torus connectivity architectures.

DiCon says both switches use the same DiCon 3D MEMS platform as its field-deployed 384×384 Optical Matrix Switches and “carry the same proven reliability.” The company positioned the 300×300 and 64×64 as lower-cost options by removing “premium features” and “many advanced options,” with the stated goal of reducing cost and passing savings to end customers. For optical budget planning, DiCon lists end-to-end loss through the OCS at less than 3.0 dB, which it says meets requirements set by hyperscalers and data center operators.

OCS products like these are typically used to create reconfigurable, all-optical paths without O-E-O conversion, which can matter when operators are trying to manage cabling complexity and connectivity patterns across large GPU fabrics.

Alongside the two new OCS models, DiCon also outlined additional products on its roadmap. The company described a 1024×1024 All-Optical Matrix Switch “scheduled for delivery in 2027,” targeting AI data centers, quantum computing, cybersurveillance, and test automation. DiCon also announced a “384 Anyport-to-Anyport Flexible Matrix Switch,” which it says is designed for high-volume co-packaged optics (CPO) production and general photonic device testing. DiCon said this platform can be dynamically reconfigured, giving an example range “from 1×383 to 192×192,” to implement new test plans in changing test environments.

On manufacturing scale, DiCon said it has “aggressively scaled its production capacity” and expects to deliver “over 3,000 matrix switch units across various configurations within this calendar year alone,” with plans to “multiply its output by many folds in the coming years.”

“By delivering field-proven performance at a competitive price point, we are ensuring that the next generation of AI and quantum scale-out can happen reliably and efficiently,” said Jeff Lee, DiCon’s Chief Product Officer.

Source: DiCon Fiberoptics

Get Data Center Engineering News In Your Inbox:

Popular Posts:

Elvis-Leka,-New-Product-Development-Engineer-—-Parker,-Sporlan-Division
From air to two-phase liquid: how rack cooling options compare on density and risk
Delta-Electronics-Americas
Delta demos 800 VDC AI data center power racks and 2.4MW CDUs at GTC
Molex Incorporated CPO-Solution
Molex debuts high-radix optical circuit switch for AI cluster fabrics
TI-800-VDC-power-architecture
TI shows 800 VDC power architecture for NVIDIA AI data centers
Screenshot
Five AI data centers to reach 1 GW power capacity in 2026, new analysis shows

Share Your Data Center Engineering News

Do you have a new product announcement, webinar, whitepaper, or article topic? 

Get Data Center Engineering News In Your Inbox: