Beyond Copper: Fabric.AI and Kopin Model Up to 60% Increase in GPU Utilization in a Neural I/o™-Powered Data Center

Initial link testing has begun, with a public demonstration targeted for January 2027 and first-generation production silicon for 2028.

Rhea-AI Summary

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.

Positive

• Minor pointNeural I/o development has advanced to initial transmission-link testing with Kopin.
• Minor point. Forward-looking: it has not happened yet and may not happen.Public technology demonstration is targeted for January 2027.
• Minor point. Forward-looking: it has not happened yet and may not happen.First-generation production silicon is targeted for 2028.

Negative

• Minor pointFuture-system performance figures are modeled results, not measured performance from production systems.
• Minor point. Forward-looking: it has not happened yet and may not happen.Implementation requires compatible switch silicon and GPU-package access through the open hardware ecosystem.
• Minor point. Forward-looking: it has not happened yet and may not happen.Technical requirements include coded-channel speed characterization, field-stable micron-level optical alignment and lower-loss glass for building-scale reach.
• Minor point. Forward-looking: it has not happened yet and may not happen.Development also requires execution ahead of competing approaches.

Argus 15 min delay 5 alerts

+2.58% vs previous close $3.38 last price 1.1x rel. volume Open Argus

Details

Market Reaction – KOPN

$3.31 – $3.49 Day Range

$609.59M Mkt Cap at Last Price

On Oct 19, the day this news came out, the latest delayed price for KOPN is 2.58% above the previous close. Our momentum scanner has recorded 5 alerts for this stock so far that day. The latest delayed price is $3.38.

Data tracked by StockTitan Argus (15 min delayed). Upgrade to Gold for real-time data.

Key Figures

Training utilization: Approximately 50% to 60% Useful training throughput per GPU: 1.3 to 1.6 times more Training run time reduction: 23% to 38% less time +4 more

Training utilization

Approximately 50% to 60%

Modeled for a 10,000-GPU system, versus typical 35% to 45% for dense models today

Useful training throughput per GPU

1.3 to 1.6 times more

Modeled Neural I/o result

Training run time reduction

23% to 38% less time

Modeled for the same training run

Interconnect energy reduction

Roughly 75% to 83%

Modeled design target

One-hop compute domain

72 GPUs to roughly 576 to 1,152 GPUs

Modeled expansion from a single rack to a row

Public technology demonstration

January 2027

Targeted milestone

First-generation production silicon

2028

Targeted production timing

Previous AI Reports

3 past events · Latest: Oct 06

Same Type 3 events

1. Oct 06

   Commercialization planning

   24h Move

   +0.2%

   Kearney engaged for Neural I/o commercialization planning; technical discussions were underway.

2. Sep 23

   Patent applications

   24h Move

   -1.4%

   Two Neural I/o patent applications covered fiber-coupled and connector-free optical designs.

3. Aug 18

   Development committee

   24h Move

   -5.3%

   Fabric.AI and Kopin launched a joint committee overseeing Neural I/o development and commercialization planning.

24h Move is the share-price change in the day after each event; other market factors may also have contributed.

Key Terms

serdes, erasure coding, telemetry

3 terms

serdes technical

"approximately one picojoule per bit compared with four to six for copper and SerDes"

SerDes (short for serializer/deserializer) is an electronic function that converts parallel streams of data into a single fast serial stream and then converts it back, like packing many lanes of traffic into one high-speed highway and unpacking them at the other end. Investors care because SerDes chips and blocks determine how quickly and efficiently devices, data centers and networks can move information; improvements or bottlenecks affect product performance, production costs, and demand across the semiconductor and communications supply chain.

erasure coding technical

"Erasure coding across the array lets the link absorb individual channel failures silently"

A data protection method that breaks a file into pieces, adds extra redundant fragments, and spreads those pieces across multiple drives or locations so the original file can be rebuilt even if some fragments are lost. Think of it like cutting a document into many strips, making a few extra patterned strips, and storing them in different safes so you can reconstruct the document if some safes are lost. For investors, erasure coding matters because it affects storage durability, recovery speed, and cost efficiency for firms that manage large amounts of digital data.

telemetry technical

"carry control and telemetry traffic"

Telemetry is the automatic collection and transmission of measurements from remote devices, systems, or patients to a central system for monitoring and analysis—like a car sending engine, speed and location data back to a dashboard. For investors it matters because telemetry provides real-time evidence of product performance, safety and user behavior, helping assess revenue potential, operational risk, regulatory compliance and whether a product is meeting market demand.

AI-generated analysis. How Rhea-AI works. Not financial advice.

See more from StockTitan in Google Search and AI answers. Adds StockTitan as a preferred source · opens Google

Add on Google Not now

Joint white paper finds today's AI GPUs sit idle half to two-thirds of the time due to copper interconnect limits; MicroLED optical design modeled to cut interconnect energy ~80% and expand the one-hop AI pod from 72 GPUs to ~1,000, with a public demonstration slated for January 2027

NEW YORK, NY, Oct. 19, 2026 (GLOBE NEWSWIRE) -- Fabric.AI, Inc. (Nasdaq: FABC) (“Fabric.AI” or the “Company”), a fabless semiconductor company developing MicroLED-based optical interconnects for AI infrastructure, today released a white paper with Kopin Corporation (Nasdaq: KOPN) that imagines how the copper wires inside AI data centers can be replaced with massively parallel MicroLED optical channels. The paper models that Neural I/o™, the MicroLED optical interconnect the two companies are jointly developing, could let a 10,000-GPU system deliver 1.3 to 1.6 times more useful training throughput per GPU, finishing the same training run in 23% to 38% less time, because the interconnect’s spare channels connect, synchronize and monitor the system instead of leaving GPUs waiting.

The white paper, “The Data Center in a SerDes-Free World,” is authored by Kopin Chief Executive Officer Michael Murray, and Fabric.AI Senior Strategic Advisor James Altucher. It looks past the performance of a single link to ask a system-level question: if the interconnect stops being the constraint, how should an AI data center be built?

The white paper can be found on Fabric A’s site here.

Today, the answer is set by copper, and copper can’t argue with physics. At 200 gigabits per second, a copper link reaches about one meter, and that meter caps a tightly connected AI compute domain at 72 GPUs in a single rack. A 10,000-GPU cluster is therefore roughly 139 separate domains joined by a slower network, and traffic that leaves a rack faces a four- to eight-fold drop in bandwidth. The paper cites typical model FLOPs utilization of 35% to 45% for frontier-scale dense training and 20% to 35% for mixture-of-experts training, meaning the most expensive machines in the world sit idle for half to two-thirds of their cycles, with a meaningful share of that loss traced to the interconnect.

“Every AI data center in the world is designed around how far copper can carry a signal, and that distance is about one meter,” said James Altucher, Senior Strategic Advisor to Fabric.AI. “The rack, the cooling, the density and the 72-GPU ceiling all follow from it. Fabric.AI was founded to remove that constraint. This paper is our system-level view of what replaces it: a row of a thousand GPUs that behaves like one machine, and an interconnect that does far more than move data.”

Neural I/o takes the opposite approach to the industry’s race for faster serial lanes. Instead of a few channels at extreme speed, it uses thousands of channels at modest speed: 2,025 physical optical channels at two gigabits per second each in a die about one millimeter square, with 1,600 channels carrying 3.2 terabits per second of payload and 425 reserved for redundancy and sideband functions. The link is built from MicroLEDs, image-sensor-derived receivers and molded plastic optics, is designed to run at approximately one picojoule per bit compared with four to six for copper and SerDes, and extends high-bandwidth reach from about one meter to tens of meters.

“The most interesting result isn’t raw bandwidth,” continued Mr. Altucher. “When you have 2,025 channels instead of a handful, the spare ones stop being insurance and start doing work. They fix errors, keep a thousand GPUs in step and watch the health of the system while it runs. Copper can’t do that, because copper has no spare channels to give.”

Key findings include:

• More work from the same GPUs. For a 10,000-GPU system, the architecture is modeled to raise training utilization from a typical 35% to 45% for dense models today to approximately 50% to 60%, delivering 1.3 to 1.6 times more useful training throughput per GPU. The gains come from reducing communication, synchronization and reliability losses, not from faster GPUs.
• The pod becomes the row. With reach measured in tens of meters, switches can move out of compute racks to the center of a row, and the one-hop compute domain can grow from 72 GPUs to roughly 576 to 1,152 GPUs. The copper cable wall at the back of today’s racks gives way to sparse fiber runs and free-space optical links between trays.
• An interconnect that helps run the machine. The 425 spare channels, provisioned for manufacturing yield, become a platform. Erasure coding across the array lets the link absorb individual channel failures silently, with no detection event or interruption. A dedicated optical barrier network is modeled to synchronize a 1,000-GPU row in well under a microsecond, deterministically, removing a synchronization cost that otherwise grows with cluster size. The same channels can distribute precise timing, carry control and telemetry traffic, and let the fabric monitor its own health.
• Lower energy and more freedom in where AI can be built. At approximately one picojoule per bit, the design targets the removal of roughly 75% to 83% of interconnect energy, along with the heat the cooling plant would otherwise remove. Because reach no longer dictates density, operators could spread compute out, extend air cooling and deploy in existing buildings that cannot host 120-kilowatt liquid-cooled racks.
• A path toward building-scale AI. The paper’s longer-term “Cube” concept models approximately 100,000 GPUs organized as about 100 rows of roughly 1,000 GPUs, connected through an optical spine with cross-building latency of about one microsecond. The paper models that the 1.3 to 1.6 times per-GPU gain can hold at that scale, where today’s utilization losses normally grow.

The paper is direct about what must go right, naming six requirements: switch silicon that accepts thousands of parallel optical channels natively, access to GPU and accelerator packages through the open hardware ecosystem, characterization of coded per-channel speed, micron-level optical alignment that holds up in the field, lower-loss glass for building-scale reach, and execution ahead of competing approaches.

Fabric.AI and Kopin have begun initial transmission-link testing of Neural I/o, with a public technology demonstration targeted for January 2027 and first-generation production silicon targeted for 2028.

Performance figures for systems that do not yet exist, including training throughput, utilization, latency, energy per bit and future system architectures, are modeled results based on assumptions detailed in the white paper and are not measured performance results from production systems.

The full white paper, “The Data Center in a SerDes-Free World: From the Copper Wall to the Row — and From the Row to One Machine the Size of a Building,” is available on Kopin’s website at: https://www.kopin.com/the-data-center-in-a-serdes-free-world-lander/.

About Fabric.AI

Fabric.AI, Inc. (Nasdaq: FABC) is a fabless semiconductor company developing MicroLED-based optical interconnects designed to address the data-movement bottleneck in AI data centers. Its Neural I/o™ program is being developed in partnership with Kopin Corporation. For more information, visit www.fabricai.com.

About Kopin Corporation

Kopin Corporation (Nasdaq: KOPN) is a leading developer and provider of innovative display, Optical Interconnect devices and application-specific optical solutions sold as critical components and subassemblies for defense, enterprise, professional and consumer products. Kopin’s portfolio includes microdisplays, display modules, eyepiece assemblies, image projection modules and vehicle mounted and head-mounted display systems that incorporate ultra-small high-resolution Active Matrix Liquid Crystal displays (AMLCD), Ferroelectric Liquid Crystal on Silicon (FLCoS) displays, MicroLED displays (µLED) and Organic Light Emitting Diode (OLED) displays, a variety of optics and low-power ASICs and optical interconnect devices for data centers. For more information, please visit Kopin’s website at www.kopin.com. Kopin is a trademark of Kopin Corporation.

Forward-Looking Statements

This press release contains “forward-looking statements” within the meaning of the federal securities laws, including statements regarding the development, capabilities and performance of the Neural I/o™ interconnect; modeled training throughput, utilization, latency, energy efficiency and system architectures described in the white paper; the anticipated timing and outcome of the planned Neural I/o public demonstration and production silicon; the Company’s partnership with Kopin; and the Company’s strategic priorities and competitive positioning. These statements are based on current expectations and assumptions and are subject to significant risks and uncertainties that could cause actual results to differ materially, including the risk that the Neural I/o demonstration or production silicon is delayed or unsuccessful; that the technology does not achieve the bandwidth, energy-efficiency, latency, reliability or cost advantages it targets; that modeled system-level results described in the white paper are not realized in practice; that switch silicon, packaging, optical alignment or fiber technologies required by the architecture are not developed or adopted; that the Company’s pending patent applications do not result in issued patents or in meaningful protection; that competing technologies achieve similar or better results first; and that the Company requires additional capital to fund its operations. Performance figures referenced herein for systems not yet in production are design targets or architectural models, not measured results of shipping products. A discussion of these and other factors is set forth in the Company’s most recent Annual Report on Form 10-K and subsequent reports filed with the Securities and Exchange Commission. The Company undertakes no obligation to update any forward-looking statement except as required by law.

Neural I/o™ is a trademark of its owner(s). All other trademarks are the property of their respective owners.

Contacts

IR Contact:
CORE IR
212-644-0924
ir@fabric-ai.co

Media Contact:
Fabric.AI
press@fabric-ai.co
www.fabricai.com 

FAQ

AI-generated questions and answers. How Rhea-AI works. Not financial advice.

What GPU utilization gains do Fabric.AI and Kopin model for Neural I/o?

For a 10,000-GPU system, the companies model dense-training utilization increasing from a typical 35% to 45% today to approximately 50% to 60%. They model 1.3 to 1.6 times more useful training throughput per GPU. These are assumption-based projections, not measured production-system results.

When do Fabric.AI and Kopin plan to demonstrate and produce Neural I/o?

A public technology demonstration is targeted for January 2027, and first-generation production silicon is targeted for 2028. The companies have begun initial transmission-link testing; neither target represents a completed milestone.

What is the building-scale Neural I/o concept described by Fabric.AI and Kopin?

The longer-term “Cube” concept models approximately 100,000 GPUs arranged in about 100 rows of roughly 1,000 GPUs, connected through an optical backbone. The companies model cross-building latency of about one microsecond and a 1.3 to 1.6 times per-GPU throughput gain at that scale. This is a modeled architecture, not an operating system.