TLDR: Eliyan Corporation is transforming AI chip design with its NuLink and NuLink-X PHY solutions, which offer high-bandwidth, low-power, and flexible chip-to-chip and die-to-die connectivity on standard organic substrates. This innovation directly addresses the ‘memory wall’ bottleneck in generative AI, enabling more powerful, efficient, and cost-effective multi-chiplet architectures. By reducing reliance on expensive silicon interposers and proprietary advanced packaging, Eliyan democratizes chiplet technology and accelerates the development of next-generation AI and HPC solutions.
Eliyan Corporation is spearheading a significant transformation in artificial intelligence (AI) chip design with its groundbreaking modular interconnect technology, specifically its NuLink and NuLink-X Physical Layer (PHY) solutions. This innovation directly addresses the long-standing ‘memory wall’ problem and critical bottlenecks inherent in current generative AI systems, promising a new era of performance, efficiency, and design flexibility for AI hardware.
The explosive growth of generative AI and large language models (LLMs) has placed unprecedented and escalating demands on computational resources and high-bandwidth memory. Traditional monolithic chip designs and expensive silicon interposers have struggled to keep pace with the need for efficient data movement and increased memory capacity. Eliyan’s technology offers a compelling alternative by enabling advanced-packaging-like performance on more accessible and cost-effective standard organic substrates.
At the core of Eliyan’s offering are its NuLink and NuLink-X PHY solutions, which reinvent chip-to-chip and die-to-die connectivity. These modular semiconductor technologies deliver unprecedented bandwidth, significantly lower power consumption, and enhanced design flexibility. By achieving high-performance interconnectivity without the limitations and expense of traditional silicon interposers, Eliyan dramatically increases the memory capacity and performance of High-Bandwidth Memory (HBM)-equipped GPUs and ASICs—the foundational components of modern AI infrastructure.
This approach not only reduces the overall cost and complexity associated with high-performance multi-chiplet designs but also democratizes chiplet technology, making it more accessible to a broader range of companies. This move lessens the industry’s dependence on proprietary advanced packaging monopolies, such as TSMC’s CoWoS, fostering greater innovation in high-performance AI and High-Performance Computing (HPC) solutions.
Power efficiency is another critical aspect of Eliyan’s innovation, offering vital benefits for energy-intensive AI data centers and contributing to more sustainable AI development. By tackling the pervasive ‘memory wall’ and the inherent limitations of monolithic chip designs, Eliyan is poised to accelerate the development of more powerful, efficient, and economically viable AI systems.
The company’s technology is set to create significant ripples across the semiconductor and AI industries, offering profound benefits and competitive shifts. AI chip designers, including industry giants like NVIDIA Corporation (NASDAQ: NVDA), Intel Corporation (NASDAQ: INTC), and Advanced Micro Devices (NASDAQ: AMD), stand to gain immensely from these advancements, enabling them to develop next-generation AI hardware that can meet the escalating demands of the AI era. Eliyan’s proven IP offers an accelerated path to market and immediate performance gains for these major players, even as they pursue internal interconnect efforts.
Eliyan’s ongoing contributions to open standards, such as the Bunch of Wires (BoW) 2.1 enhancements and the newly announced BoW Memory Addendum within the Open Compute Project (OCP), further solidify its leadership. These efforts define optional enhancements for connecting ASICs directly to memory devices through BoW interfaces, allowing designers to achieve higher performance and power efficiency for memory-intensive workloads like AI training and inference. The company recently showcased these innovations, including a 64Gbps SBD demo in TSMC’s 3nm process technology, at the OCP Global Summit 2025.
Also Read:
- Rebellions and Arm Partner to Advance Next-Generation AI Data Center Infrastructure
- Vertical Semiconductor Secures $11M Seed Funding to Advance Vertical GaN Chips for AI Power Efficiency
This foundational shift in AI hardware design and scaling, seamlessly integrating with and accelerating the broader trends of chiplets and generative AI, positions Eliyan as a key enabler for the future of high-performance, energy-efficient computing.
