#Taiwan #Taipei #Neuchips #Vecow #GSH

Neuchips Collaborates with Vecow and GSH to Revolutionize Data Processing with Offline Generative AI

In anticipation of the Embedded World 2025 event, Neuchips, a prominent supplier of application-specific integrated circuits (ASICs) focused on artificial intelligence (AI), has announced a strategic partnership with Vecow and Golden Smart Home (GSH) Technology Corp. This team-up aims to innovate SQL data processing through a private, secure, and energy-efficient AI solution, enabling real-time insights from internal databases via natural language queries.

Ken Lau, CEO of Neuchips, stated, "Our collaboration with Vecow and GSH exemplifies the future of industrial AI deployment. At Embedded World 2025, visitors will see how our Viper AI acceleration card, supporting models with 12 billion parameters while only consuming 45W, perfectly complements Vecow's robust Edge AI computing systems and GSH's ShareGuru solutions. This powerful synergy allows for secure and efficient proprietary data processing that meets the demands of modern industrial environments. We are proud to partner with Vecow in bringing this generative AI innovation to enterprise-focused applications."

As generative on-site AI applications expand, Joseph Huang, Executive Vice President at Vecow, highlighted the rapidly growing demand for multimodal large language models (LLMs). He noted, "As a provider of edge AI computing solutions, Vecow is excited to collaborate with Neuchips to develop cutting-edge LLM solutions based on Retrieval-Augmented Generation (RAG). This partnership allows users access to the latest data without training models, yielding more relevant and higher-quality results. Our clientele is in search of compact, cost-effective, low-energy AI workstations that outperform traditional cloud-based GPU solutions."

Addressing Complex Data Processing Needs

The increasing complexity of databases, combined with the scarcity of SQL skills, has significantly hampered the extraction of vital information. However, online AI models cannot be employed due to a lack of protection for proprietary information. To tackle these issues across diverse industries, this groundbreaking solution leverages the Vecow ECX-3100 RAG Edge AI Inference workstation. This LLM-compatible RAG computing platform operates the GSH-powered ShareGuru QA 2.0 solution using a single Neuchips LLM card - the Viper AI series card.

This synergy empowers users to translate human language into SQL queries, making these queries more accessible and efficient while simultaneously lowering SQL expertise costs. Moreover, the solution provides maximum data privacy and security since the Neuchips offline card facilitates local execution of the solution and ShareGuru platform, ensuring high precision through AI-driven query validation.

The Viper series from Neuchips demonstrates high energy efficiency, running at just 45W with a full model featuring 12 billion parameters. Introduced during COMPUTEX 2024, the Viper series frees over 90% of the resources needed for generative AI from the processor, unlocking the full potential of LLMs with the following standout features:

• - Additional memory capacity of 32GB
• - Native support for structured language models in BF16
• - The launch of the Raptor generative AI accelerator chip at CES 2024

Looking ahead to 2026, Neuchips aims to enhance performance with low-power multimodal ASICs.

About Neuchips

As a leader in ASIC solutions for AI, Neuchips is at the forefront of developing hardware specifically designed for DLRM and LLM applications. With a dedicated team of experts, commitment to innovation, and a strong presence in industrial organizations, the company is poised to continue shaping the future of AI hardware and ushering in a new era of efficiency and performance. To learn more, visit Neuchips.