GSI Technology, Inc. Q2 FY2026 earnings call

• Announced a Cornell University paper verifying Gemini-I chip performed on par with NVIDIA's A6000 on certain AI tasks with 98% less energy; Gemini-II has 8x memory and 10x performance of Gemini-I. • Closed a $50 million equity financing to accelerate hardware and software build-out. • Begun work to acquire IP for Plato to start hardware development and building software team for Plato. • Investing in Gemini-II software solutions and applications to make platform accessible for developers, especially in edge and defense applications. • Advancing SBIR and POC engagements; Gemini-II approved for prototyping by offshore defense contractor for SAR applications, demonstrating low power and performance; joint POC with defense orgs and drone partner showed Gemini-II outperforms competitors in time to first token. • Plato program aims for tape-out in early 2027, engaging strategic partners for funding, testing, and prototyping; enhances military defense functions like SAR imaging, object recognition, etc. • Received $751,000 extension of Space Development Agency contract for radiation hardened beam testing of Gemini-II for aerospace applications.

In the second quarter of fiscal 2026, sales to KYEC were $802,000 or 12.5% of revenues. Sales to Nokia were $200,000 or 3.1% of revenues. Sales to Cadence Design Systems were $1.4 million or 21.6% of net revenues. Military defense sales were 28.9% of second quarter shipments. SigmaQuad sales were 50.1% of second quarter shipments in fiscal 2026.

• For the third quarter of fiscal 2026, expect net revenues in the range of $6.0 million to $6.8 million with gross margin of approximately 54% to 56%. • Focused on disciplined execution to bring Gemini-II to market, advance Plato road map, and drive long-term shareholder value.

Q: Congratulated on the Cornell verification and asked if they've worked with auto industry on autonomous vehicles?

A: Have not yet, focused on military defense arena now but believe technology will adapt to auto industry in future.

Q: Asked about $50 million placement, if with strategic investor, holding period?

A: Not with strategic investor, no required holding period for shares.

Q: Asked if there have been strategic inquiries since Cornell report?

A: Looking at things and talking to parties, but no specific significant new inquiries yet.

Q: Asked if there's a plan to have third party validate Gemini-II?

A: Yes, talking to Cornell and other researchers to get Gemini-II board for validation.

Q: Asked about military defense management/board experience plans?

A: No current plans to revise Board, but could in future if makes sense.

Q: Asked why develop own LLM vs using existing; A: POC requirements, model supports multi-modal well like image and text as requested by partners.

Q: Asked how differentiating APU vs GPU in power, latency, cost efficiency?

A: Cornell paper showed 98% less power than NVIDIA GPU; SAR algorithm has faster image creation at lower power; beats competitors on power and sometimes performance.

Q: Asked about engagement/partnership discussions with larger semiconductor/AI companies?

A: Focused on current customers, no recent discussions with other semiconductor companies.

Q: Asked about power factors in AI data centers?

A: Focused on edge now, Gemini-II and Plato are for edge, with low power like 15 watts for Gemini-II and 4-15 watts for Plato.

Q: Asked about total addressable market and ramp, pricing?

A: Drone market is large (tens of billions), pricing varies by market, boards could be a few thousand to $10,000, chips $1,000 or more.

Q: Asked about capital deployment of $50 million raise among Gemini-II, software, Plato?

A: Plato has fixed costs like IP and mass tape-out, rest evenly split between Gemini-II and Plato.

Q: Asked about concrete milestones and dates for Gemini-II pilot shipments, purchase orders/LOIs?

A: Plan pilot shipments in first half of 2026, customers still in evaluation, but offshore defense contractor has good acceptance and looking for practical applications.

Q: Asked about software stack maturity and access for developers?

A: Developing library and algorithm for Gemini-II, then moving to tool and compiler work with partners and customers.

Q: Asked why not going after data centers?

A: Previous data center product needed different partner and more funding, too expensive, focusing on edge now.

Q: Asked about where $50 million raise went and future capital allocations?

A: Transaction closed after balance sheet date, allocating to IP for Plato and splitting between software and Plato development.

Q: Asked if attracted more interest from potential new customers after Cornell study?

A: Customers aware of low-power advantage from previous talks, Cornell paper was third-party validation of what they already knew.