Omni Design's 3nm Optics IP Aims to Break AI's Power Bottleneck

SAN JOSE, CA – February 06, 2026 – As artificial intelligence continues its exponential growth, the data centers that power it are straining under an unprecedented challenge: an impending power and bandwidth crisis. To address this critical bottleneck, Omni Design Technologies has announced a significant advancement in its Co-Packaged Optics (CPO) portfolio, introducing an analog front-end (AFE) intellectual property (IP) core built on a cutting-edge 3nm process. This development promises to dramatically improve the efficiency of moving data at 200G-class speeds, a crucial step in enabling the next generation of hyperscale AI infrastructure.

The new IP, which includes a high-swing linear transmit (TX) driver and a low-noise receive (RX) transimpedance amplifier (TIA), is designed to be the vital link between electrical silicon and photonic components. By placing optical I/O closer to the central processing silicon, Omni Design aims to slash power consumption and increase bandwidth density, tackling the two biggest hurdles to scaling AI systems sustainably.

The Inevitable Shift to Co-Packaged Optics

The insatiable appetite of large language models and other AI workloads for data has turned modern AI clusters into vast, distributed supercomputers. The connections between the thousands of processors in these clusters are paramount, but traditional copper wiring and even pluggable optical modules are hitting a wall. At speeds of 200 Gb/s and beyond, electrical signals traveling over copper degrade quickly, requiring power-hungry digital signal processors (DSPs) and retimers to clean them up. This process generates significant heat and consumes a disproportionate amount of energy, a critical concern as data centers are projected to consume a rapidly growing share of the world's electricity.

Co-Packaged Optics represents a fundamental architectural solution to this problem. By integrating optical components directly onto the same package as the switch or compute ASIC, CPO drastically shortens the electrical path data must travel. This proximity minimizes signal degradation, reduces the need for power-intensive signal conditioning, and allows for a much higher density of I/O connections. The industry has taken notice, with market forecasts projecting the CPO market to surge from tens of millions today to over $2.6 billion by 2033, according to analysts at Yole Group. Major players like NVIDIA, Broadcom, and Intel are all heavily invested in CPO, viewing it as a foundational technology for future data centers.

The 3nm Frontier: A Double-Edged Sword for Analog Design

Omni Design's decision to develop its CPO AFE on an advanced 3nm process node is a strategic move that places it at the bleeding edge of semiconductor technology. Migrating to smaller nodes like 3nm offers tremendous advantages, primarily by enabling higher transistor density and lower power consumption. For optical interconnects, this translates directly to a lower energy-per-bit metric, which is the gold standard for data center efficiency. The advanced process also allows for faster transistors, which is essential for handling the high-frequency signals of 224Gb/s PAM4 modulation used in 200G-class links.

However, this frontier is not without its perils. Analog circuit design, which forms the heart of an AFE, becomes exponentially more difficult on advanced process nodes. Transistors behave less ideally, suffering from lower intrinsic gain, increased signal noise, and greater variability from manufacturing imperfections. Designing a high-performance linear driver and a low-noise amplifier under these conditions requires deep, specialized expertise. This is where a dedicated IP provider like Omni Design finds its niche. By mastering the complexities of analog design on 3nm, the company provides a critical building block that enables larger system-on-chip (SoC) developers to integrate high-speed optical capabilities without having to develop the complex analog expertise in-house.

Powering the AI Data Center of Tomorrow

The new AFE IP is the engine that drives the electrical-to-optical conversion. The transmit driver takes high-speed electrical data from the switch ASIC and converts it into a precise analog signal to drive the laser, while the transimpedance amplifier on the receiving end takes the faint signal from a photodetector and amplifies it into a clean electrical signal. The performance of these two components is critical to the integrity and reliability of the entire optical link.

“As AI training and inference fabrics grow, the industry is being forced to rethink power and bandwidth at every interface. Co-Packaged Optics is a critical architectural shift, with the analog front-end playing a key role in achieving system-level performance and efficiency,” said Kush Gulati, President and CEO of Omni Design Technologies, in the company's announcement. “This development continues to leverage our deep analog and mixed-signal expertise to help customers push optical I/O closer to the compute and switching silicon—improving bandwidth density while driving down power.”

By supporting 224Gb/s PAM4 signaling, Omni Design's IP directly addresses the throughput demands of next-generation AI scale-up and scale-out fabrics, ensuring that data can move between processors without creating a system-level bottleneck.

A Focused Strategy in a Competitive Landscape

While Omni Design is not building the entire CPO-enabled switch, its strategy as a specialized IP provider positions it as a crucial enabler within a fiercely competitive ecosystem. The company's focus on high-performance, ultra-low power “Wideband Signal Processing” solutions extends beyond CPO. Its portfolio includes world-class data converters and IP cores that have been adopted by tier-one customers in demanding markets such as automotive radar, 5G/6G wireless, and satellite communications. This broad base of silicon-proven technology and a recent, oversubscribed Series A funding round of over $35 million underscore market confidence in the company's capabilities.

This CPO advancement is a logical extension of that core competency, applying its proven expertise in high-speed analog and mixed-signal design to one of the most pressing problems in the tech industry today. By providing these foundational IP building blocks, Omni Design allows the titans of the AI hardware world to accelerate their own roadmaps. This symbiotic relationship highlights the often-unsung role of IP providers in driving the entire semiconductor industry forward, proving that the most impactful innovations are not always the finished product, but the enabling technologies that make them possible.