GSI (GSIT) 16th Annual LD Micro Invitational Conference summary

Company overview and strategic direction

• Focused on transitioning from SRAM memory to AI-centric products, leveraging 30 years of expertise and a strong partnership with TSMC.

• Raised $47 million in October to support AI product development, with $67 million in cash and no debt.

• Maintains a lean workforce of 126, primarily engineers, and holds 147 patents, 88 of which are for the APU.

• Insider ownership exceeds 20%, with original angel investors still holding shares.

• SRAM business remains profitable, funding AI R&D and maintaining a generational lead as competitors have frozen their roadmaps.

AI product line and technology differentiation

• Developed the Associative Processing Unit (APU), a true compute-in-memory (CIM) solution, minimizing data movement and power consumption.

• Focused on edge applications rather than data centers, targeting a $10 billion market with 20% CAGR.

• Demonstrated 98% lower power consumption than NVIDIA GPUs in Cornell's RAG application study.

• Won competitive bake-offs for drone surveillance, outperforming NVIDIA Jetson and Qualcomm Snapdragon in both power and latency.

• Gemini-II chip powers current POCs, including a Department of Defense drone project and a smart city deployment in Taiwan.

Market applications and customer wins

• Edge AI applications include autonomous drones, satellite SAR imagery, and smart city surveillance.

• Smart city POC in Taiwan progressing through three phases, potentially scaling to 6,000 cameras by next year.

• US Army awarded a $2 million SBIR phase II grant for ruggedized edge nodes, with commercialization potential.

• Space Development Agency testing APU for radiation tolerance in satellite applications.

• Actively pursuing larger non-dilutive funding opportunities, including STRATFI and BAA grants.