Вештачка интелигенција

Introducing Gemini-II: A New Processor for Artificial Intelligence and Aerospace Industries

Summary

GSI Technology Inc. is making waves in the world of artificial intelligence and aerospace with its new Gemini-II processor. The California-based company has gained significant attention for its innovative technology, which is poised to revolutionize industries like drones, satellites, and […]

Introducing Gemini-II: A New Processor for Artificial Intelligence and Aerospace Industries

GSI Technology Inc. is making waves in the world of artificial intelligence and aerospace with its new Gemini-II processor. The California-based company has gained significant attention for its innovative technology, which is poised to revolutionize industries like drones, satellites, and airplanes.

With its headquarters in Sunnyvale, California, GSI Technology has been at the forefront of designing and developing memory products for the telecommunications and networking industries since its establishment in 1995. While they are known as a leading supplier of static random-access memory (SRAM) semiconductor products, it is their advancements in AI processor technology that have caught the industry’s attention.

GSI recently announced a major milestone in the development of the Gemini-II processor, designed specifically for the artificial intelligence market. The company revealed the completion of the design for the Gemini-II’s associative processing unit, highlighting significant performance improvements over its predecessor, Gemini-I. The new processor offers over 10 times better processing performance and eight times higher memory density.

The 16-nanometer Gemini-II chip is set to be available to the public in 2024. Didier Lasserre, Vice President of Investor Relations at GSI Technology, stated that while the 28-nanometer Gemini-I processor is excellent for applications like synthetic imaging radars or molecular research, Gemini-II will expand the capabilities of this technology to “real applications in data centers.”

GSI Technology is already in discussions with a leading cloud service provider about Gemini-II. “Large language models like Microsoft’s Bing and Google’s BERT are pushing the boundaries of natural language processing,” said Li-Lian Shu, CEO of GSI Technology. “We are only beginning to explore their potential for transformation across various industries and applications. Gemini-II and future AI processor implementations open up numerous possibilities.”

Lasserre added that Gemini-II could also be utilized in driver assistance systems in the automotive industry, opening up new markets for the company.

FAQ

What is the Gemini-II processor?

The Gemini-II processor is a groundbreaking technology developed by GSI Technology Inc. It is designed for applications in artificial intelligence and aerospace industries, offering significant performance improvements over its predecessor.

When will the Gemini-II chip be available?

The 16-nanometer Gemini-II chip is scheduled to be released to the public in 2024.

What industries can benefit from Gemini-II?

Gemini-II has applications in various industries, including artificial intelligence, aerospace, data centers, and driver assistance systems in the automotive industry.

What are the advantages of Gemini-II over its predecessor?

Gemini-II offers over 10 times better processing performance and eight times higher memory density compared to Gemini-I.

While companies like Nvidia Corp., Palantir Technologies Inc., and Microsoft Corp. have become well-known in the field of artificial intelligence, lesser-known AI companies are also attracting investor attention this year.

GSI Technology has seen astronomical growth in its stocks, with a 33.5% increase this year, surpassing the S&P 500’s 10.1% growth.

Mark Wright, Director of AI Processors at GSI Technology, emphasized that the low power consumption and high performance of Gemini-II chips make them a viable choice for the aerospace industry. He cited their ability to be directly integrated into airplanes, drones, or satellites, where they can apply AI technology to image data, eliminating the need to send the data back to Earth for processing.

He used the example of the Malaysian Airlines flight MH370, which disappeared in 2014 with 239 people on board, triggering a massive search for a vast amount of images. By processing the image data on the satellite, Gemini-II can efficiently evaluate vast oceanic areas, sending back only relevant data to Earth.

With a market capitalization of $56.7 million, GSI Technology is relatively small compared to popular stocks like AMC Entertainment Holdings Inc. and GameStop Corp., as well as industry giant Nvidia with a market capitalization over a trillion dollars.

“We truly shine when databases become big,” said Wright, noting that using a large number of graphics processing units (GPUs) in such scenarios can be energy-intensive and costly. “We are very efficient because we process at a vector level,” he added.

“We can simultaneously search through various fields because they are all vectorized; they become just numbers,” Wright explained. “We can take a vector and immediately identify the distance between that vector and what you are searching for.”

In their fiscal results for the first quarter in July, GSI Technology reported a net loss of $5.1 million, compared to a net loss of nearly $4 million in the previous quarter. The company has reported a net loss for more than twelve consecutive quarters.

As of June 30, 2023, the company had $27.7 million in cash, cash equivalents, and short-term investments, compared to $30.6 million at the same time the previous year. “We have $27.7 million in the bank, currently with no debt, so that will certainly last a while,” said Lasserre, noting that the company has also reduced costs.

Source: Adapted from James Rogers, MarketWatch