Fabric.AI (FABC) released a joint white paper with Kopin (KOPN) modeling higher AI training utilization using their Neural I/o optical interconnect.
For a 10,000-GPU system, the companies model dense-training utilization rising from 35%–45% to approximately 50%–60%, with 1.3 to 1.6 times more useful training throughput per GPU and training runs taking 23%–38% less time. The design targets roughly 75%–83% lower interconnect energy and models expanding the directly connected compute group from 72 GPUs to roughly 576–1,152 GPUs.
Initial transmission-link testing has begun, with a public demonstration targeted for January 2027 and first-generation production silicon targeted for 2028. These future-system figures are modeled, not measured production results. Implementation requires compatible switch chips, access to GPU packages, optical validation and field-stable alignment, with lower-loss glass needed for building-scale reach.
Fabric.AI (Nasdaq: FABC) will participate in a live YouTube panel on AI infrastructure on October 20, 2026. Senior Strategic Advisor James Altucher will join Kopin CEO Michael Murray and AI expert Cortney Harding for “The Trillion Dollar AI Bottleneck,” hosted by Kopin at Nasdaq MarketSite in New York City at 11:00 a.m. EDT.
The discussion will cover data center capacity, energy demands, data movement and networking constraints, including optical interconnect technologies. Viewers can submit questions during the livestream.
Fabric.AI (FABC) has engaged consulting firm Kearney to help develop the commercialization strategy for its Neural I/o optical interconnect platform.
Fabric.AI and development partner Kopin (KOPN) have executed non-disclosure agreements with several semiconductor and data center technology companies, with technical discussions underway. Kearney will help organize a commercial pipeline, refine product positioning and identify target customers and channel partners. Fabric.AI plans hardware demonstrations in early 2027 and expects active sales discussions to begin afterward.
Fabric.AI (FABC) filed two USPTO patent applications for its MicroLED-based Neural I/o optical interconnect architecture, developed in collaboration with Kopin (KOPN).
Both applications use a single optical system that images an entire micro-LED array rather than aligning channels individually, aiming to keep assembly complexity flat as channel counts rise. The first application covers a fiber-coupled design that images the array onto a multicore fiber, while the second covers a connector-free free-space link that sends parallel data across a short air gap to a detector array. Fabric.AI expects to demonstrate working Neural I/o hardware in January 2027.
Fabric.AI (FABC) has appointed Haynes and Boone as intellectual property counsel to support development and protection of Neural I/o™ technology jointly owned with Kopin (KOPN).
Haynes Boone will advise on patent strategy, technology licensing and other IP matters for the Neural I/o™ roadmap. The appointment follows Fabric.AI entering non-disclosure agreements with multiple major AI companies for technical discussions about Neural I/o™, which is a MicroLED-based optical interconnect architecture aimed at bandwidth and energy-efficiency needs in next-generation AI data centers.
Fabric.AI (Nasdaq: FABC) announced the formal launch of a Joint Steering Committee (JSC) with Kopin (Nasdaq: KOPN) to oversee development and commercialization planning for their Neural I/o™ MicroLED-based optical interconnect platform. The JSC is chaired by Kopin CEO Michael Murray and includes senior leaders from both companies.
According to Fabric.AI, the launch coincides with expanded engagement across the AI infrastructure and interconnect ecosystem, including new NDAs with prospective participants such as companies in NVIDIA’s NVLink ecosystem. Neural I/o remains on track for its first live public demonstration at CES 2027, and the companies plan to highlight the platform at OFC and SPIE Photonics West in 2027.
Fabric.AI (Nasdaq: FABC)/b) announced expanded industry engagement for its Neural I/o™ MicroLED-based optical interconnect platform, including several new non-disclosure agreements with prospective participants in the AI interconnect ecosystem. According to Fabric.AI, these NDAs include multiple companies that are part of NVIDIA's NVLink ecosystem and are entered into under the Neural I/o initiative being developed with .
The additional agreements broaden the group of industry players evaluating Neural I/o for next-generation AI infrastructure. Fabric.AI describes Neural I/o as a massively parallel MicroLED-based optical architecture aimed at moving large data volumes efficiently between GPUs, accelerators, memory and other components, with a planned technology demonstration at CES 2027.
Fabric.AI (Nasdaq: FABC) and development partner Kopin (Nasdaq: KOPN) confirmed that their Neural I/o™ MicroLED-based optical interconnect platform remains on schedule for its first public, live demonstration at CES 2027 in January. The companies report progressing from architecture to demonstrable hardware through a series of engineering, integration and internal-development milestones, ahead of the development timeline Fabric.AI outlined in May 2026.
Neural I/o™ uses a massively parallel array of MicroLEDs as its optical engine, aiming to address bandwidth, energy-per-bit, latency and thermal constraints of conventional electrical I/O and pluggable optics in AI infrastructure. Fabric.AI cites third-party research projecting multibillion-dollar growth in optical interconnect, co-packaged optics and silicon photonics markets through 2032, and believes Neural I/o™ positions it to participate in this expansion.
Fabric.AI (Nasdaq: FABC) will present at the Canaccord Genuity 46th Annual Growth Conference in Boston, Massachusetts, on Tuesday, August 11, 2026, from 3:30–3:55 p.m. ET in the Griffin/Robinson room.
Management will deliver a corporate overview covering Fabric.AI’s strategy for AI infrastructure, focusing on communications bottlenecks in AI factories, its MicroLED-based optical interconnect development, recent strategic shifts away from digital asset treasury activities, and the broader market opportunity as governments and hyperscalers emphasize high-performance, energy-efficient, and secure AI data center communications. CEO Josh Silverman and Strategic Advisor James Altucher will jointly discuss the evolution of AI infrastructure and the company’s long-term positioning, and management will be available for one-on-one investor meetings scheduled via the conference website or investor relations.
Fabric.AI (Nasdaq: FABC) and Kopin (Nasdaq: KOPN) jointly published a white paper, “Optical Interconnects as a Critical Enabler for Next-Generation AI Infrastructure,” describing their MicroLED-based Neural I/o™ optical interconnect technology for AI data centers.
The paper compares copper, laser-based optics, and MicroLED optical links, highlighting how MicroLED’s massively parallel architecture targets sub-picojoule-per-bit efficiency and a roadmap toward 20 Tb/s aggregate bandwidth by scaling channel count rather than lane speed. According to the companies, this approach addresses bandwidth, power, heat, and scalability constraints as AI clusters link tens of thousands of accelerators.