GSI Technology, Inc.

Ladies and gentlemen, thank you for standing by and welcome to the GSI Technologies 4th Quarter Fiscal 2024 Financial Results Conference Call. At this time, all participants are in a listen-only mode. Later, we will conduct a question-and-answer session and will provide instructions for those interested in entering the queue for the Q&A. Before we begin today's call, the company has requested that I read the following safe hardware statements. The matters discussed in this conference call may include forward-looking statements regarding future events and the future performance of GSI technology that involve risks and uncertainties that could cause actual results to differ materially from those anticipated. These risks and uncertainties are described in the company's form 10-K filed with the Securities and Exchange Commission. Additionally, I've been advised to advise you that this conference call is being recorded today, May 2, 2024, at the request of GSI Technology. Hosting the call today is Li-Lin Xu, the company's chairman, president, and chief executive officer. With him are Douglas Shirley, chief financial officer, and Adir Lecer, vice president of sales. I would like to turn the conference over to Mr. Xu. Please go ahead, sir.

Good afternoon, and thank you for joining us today. Let's start with some highlights from our first quarter activities. During the quarter and subsequently, we achieved significant milestones across our key initiatives. First, we launched two high-capacity, low-power 1U and 2U servers integrated with the powerful Gemini One APU designed specifically for SAR and fast vector search applications. This service can enable mobile applications such as the plane and the satellites, and will offer enterprise-level processing at age. In addition, we now have a benchmark on A and 16 leaderboards for SAR, providing crucial tools for our follow-up efforts with SAR targets over the past year. We are finalizing marketing materials and plan to begin sales promotion to target the customers by the end of second calendar year courses. Second, we made considerable progress with chips from the first spring of Gemini 2. Presently undergoing rigorous testing and debugging. The results showcase during the testing was significant for the integration of the chip onto a board. Having a chip mounted on a board has enabled comprehensive performance assessment. The chip can perform the basic functions in the processing core and is able to move data between the processing array and the L1 and L2 local memories. It's verified that the instruction can be successfully executed through the embedding processor and that the data path is also working. So far, we are pleased with the chip's result, which has been better than we anticipated. We aim to conclude the phase by the end of calendar Q2 and then proceed with the second spin. We anticipate receiving the next chip by fall and initiate benchmarking shortly after. This could allow us to begin preliminary customer sampling with the expectation of starting alpha testing with target customers before year end, along with further software development and the writing of libraries. Due to the huge model size requirement introduced by large language models, there are many efforts in the AI industry to reduce model size. loadability, binary neural network, and the bit length. Those algorithms reduce the model weight to 1 bit or 1.58 bit instead of a 16 bit and 8 bit in the present models. The reduced weight size relieves the memory storage requirement and simplifies the computation. Thus simplifying the computation is especially beneficial to APU architecture. over the traditional GPU architecture. GPU architecture are based on high resolution matrix modification for computation that is good for high resolution model size. However, BitLab and BAMN require boolean operation and integer addition for computation instead of matrix modification. APU architecture has boolean operation and integer addition as the basic building blocks. That means that the APU is well suited for this kind of operation and offers advantage for higher performance and lower power. Our plan with Gemini 2 is to target this algorithm and demonstrate to users of BMN and the BitLab that APU can fit the small model entirely into the chip for edge applications. We expect to begin this demonstration early next calendar year once the next spring of the trip becomes available. Another important development since the close of Q4 is the sale and lease back of our headquarters in Sunnyvale, California. The sale, expected to close in early June, will provide additional funding to support the finalization of Gemini 2 and other R&D projects. Last, our ongoing engagement with the hybrid skills continue to show promise. We have received great feedback on what we need to focus on as we start to discuss the design and application for Gemini 3. We are seeking a technical partner for Gemini 3. We are pleased with the progress we have made in our conversations with some potential partners who could provide technology to support the functionality of Gemini 3. with high bandwidth memory. Now, I will hand the call over to Doug, to Didier, who will discuss our business performance further. Please go ahead, Didier.

Thank you, Lilien. Starting with fiscal year 2025, I would like to highlight several strategic goals and initiatives we will work on with each of the APU generations. First, we aim to sustain our legacy sales at the current run rate. and now with two APU servers available, we're primed to pursue Gemini 1 sales, particularly focused on SAR SaaS solution. It's our intention to achieve the first sales of Gemini 1 in the second half of fiscal 2025. With Gemini 2, we are actively writing libraries to develop new applications on the edge or near edge. The second generation of our APU brings significant performance enhancements with more than 10 times the processing power with eight times the memory density compared to Gemini 1. Gemini 2 offers substantial processing capabilities being suitable for both low power data center expansion and enabling data center functions at the edge. This empowers the local execution of computational intensive tasks, increasing the edge application capabilities like advanced driver assistance systems for automobiles and HPC in delivery drones, autonomous robots, unmanned aerial vehicles, and satellites. Additionally, Gemini 2's memory can hold a small database, a potential door opener for enhanced performance in several applications. One example could be an off-the-shelf facial recognition solution, potentially in hardware with on-prem software or SAF. This fiscal year, we are highly committed to further engaging with potential Gemini 3 partners, including customers for the chip and technology partners for HBM integration. Getting partner funding for Gemini 3 development is also a key priority. In addition, I am pleased to share that we ship parts for a new prototype for a European Space Agency robotic space mission to intercept a comet. Before I conclude, I'd like to emphasize the key objectives that the GSI team and I are actively pursuing. First, securing sales for Gemini 2. Secondly, advancing Gemini, I'm sorry, securing sales for Gemini 1. Advancing Gemini 2 for its second iteration, which will enable us to commence customer sampling later this year. And then lastly, actively seeking strategic partnerships for Gemini 3. Achieving these objectives will build awareness of the APU brand improve our financial standing, and position us to enter large, high-growth markets related to AI. Let me close with the customer and product breakdown for the fourth quarter, fiscal 2024. Sales to Nokia were $694,000, or 13.5% of revenues, compared to $1.2 million, or 21.8% of revenues in the same period a year ago, and $807,000, or 15.2% of net revenues in the prior quarter. Military defense sales were 35.5% of fourth quarter shipments compared to 44.2% of shipments in the comparable period a year ago and 28.2% of shipments in the prior quarter. Semi-quad sales were 42.4% of fourth quarter shipments compared to 46.3% in the fourth quarter of fiscal 2023 and 46.9% in the prior quarter. I'd like to hand the call over to Doug. Doug, go ahead, please. Thank you, GDA.