GSI Technology

Compute-in-memory represents a significant shift in computing paradigm with proven SRAM technology that is ready for high volume deployment.

The page explains how embedding vectors enable similarity search for personalized customer recommendations, improving user experience.

The page describes the growing adoption of neural search (vector or semantic search) in production, citing examples from Google, Home Depot, and Spotify.

The page explores the need for greater explainability in large language models (LLMs) to build trust in their decision-making processes, referencing a presentation by Stanford University’s Christopher Potts.

The page examines how natural language processing (NLP) is integrated into image search, citing Google’s MUM model as an example of its use in interpreting photos.

The page highlights GSI Technology’s granted patents in the semiconductor industry.

The page lists quality information and certifications held by GSI Technology.

GSI Technology’s international distributors assist customers in selecting parts that match their design requirements.

GSI Technology’s international sales force assists customers in selecting parts that match their design requirements.

GSI Technology’s North American distributors assist customers in selecting parts that match their design requirements.

GSI Technology secures Phase I of a smart city initiative with Hsinchu County government in Taiwan, deploying its Gemini-II Associative Processing Unit (APU). The project marks the company’s first commercial smart city application of the compute-in-memory AI hardware. GSI’s APU technology aims to redefine high-performance computing through associative processing.

May 2026GSI Technology, Inc.News article

GSI Technology achieves a **3-second time-to-first-token (TTFT)** for edge multimodal large language model inference using its **Gemini-II Compute-in-Memory processor**. The benchmark highlights the processor’s ability to accelerate AI workloads with true compute-in-memory technology, targeting high-performance applications in AI and edge computing. The company, known for its **Associative Processing Unit (APU)**, claims this as a first for edge multimodal LLM performance.

Jan 2026GSI Technology, Inc.News article

GSI Technology reports first-quarter fiscal 2026 financial results for the period ending June 30, 2025. The company provides earnings updates covering its Nasdaq-listed operations.

Jul 2025GSI Technology, Inc.News article

GSI Technology wins a $1.1 million SBIR Direct-to-Phase II contract from AFWERX to demonstrate high-data computation applications using its next-generation Gemini Associative Processing Unit (APU). The APU targets AI and high-performance parallel computing for networking, telecommunications, and military sectors. The award supports development of real-world use cases for the technology.

Jan 2024GSI Technology, Inc.News article