Silicom Ltd. (SILC) Earnings

Management highlights

- Overall Business Performance * Q2 2026 results significantly outperformed management expectations, marking a continued acceleration of year-over-year revenue growth: 17% in Q4 2025, 33% in Q1 2026, and 59% in Q2 2026 * The company's multi-year strategic plan for the core business is tracking ahead of original projections, with accelerating recurring revenue momentum driving a business inflection point * The balance sheet remains very strong: as of Q2 end, total working capital and marketable securities were $107 million (~$19 per share), including $55 million in cash, cash equivalents and highly rated marketable securities, with zero debt. Inventory was increased intentionally to support strong growth and mitigate extended memory chip lead times. - Design Win Progress * Management set a 2026 full-year target of 7 to 9 new design wins, and 7 wins had already been secured by the midpoint of the year, putting the company on track to exceed the upper end of the target * Notable 2026 Q2 and early Q3 design wins include: an FPGA SmartNIC win with a European advanced encryption leader (third post-quantum cryptography (PQC) design win, expected to scale to ~$3 million annual deployment, with additional future product opportunities); the company's first-ever white-label switching design win with a Tier 1 Global Security Leader, with first production expected by end of 2026; a new custom high-speed server adapter win with an existing blue-chip customer that triples expected 2027 business from this customer to nearly $10 million * The pipeline of additional potential design wins remains broad and deep, laying the foundation for sustained growth beyond 2026 - AI Inference Business Progress * Tangible strong progress has been achieved in the AI inference market in less than 9 months, positioning the company to capitalize on the industry shift from AI training to inference and the rise of disaggregated inference architectures * Recent milestones include a secured design win and first production order from a pioneering AI inference acceleration provider, delivery of a customized AI NIC solution to a leading AI inference ASIC/infrastructure vendor for evaluation with initial deliveries pending, and development of a new bespoke inference-specific solution based on secured orders * Management views AI inference as a potential game-changer that can significantly enhance the company's long-term growth trajectory, built on the company's existing core networking and customization expertise

Guidance

- Q3 2026 revenue is guided to a range of $25 to $26 million, representing 66% year-over-year growth at the upper end of the range, marking continued acceleration from Q2 2026 - Full year 2026 revenue guidance was raised significantly to $93 to $95 million, up from the prior guidance range of $82 to $83 million. The new guidance represents over 50% year-over-year full-year growth, driven by better-than-expected core business performance and an additional $3 to $4 million in expected 2026 revenue from AI inference production orders - Management now expects the company to return to quarterly non-GAAP profitability in the second half of 2026, significantly earlier than the original anticipated timeline - Gross margin is expected to remain within the company's prior short-to-mid-term expected range of 27% to 32%, with no dramatic change expected from new AI inference business ramp

Segment performance

Silicom does not break out formal product segment financials in this call. The consolidated Q2 2026 revenue was $23.8 million, a 59% year-over-year increase from $15 million in Q2 2025. Geographically, over the trailing 12 months, North America contributed 79% of total revenue, Europe and Israel contributed 13%, and Far East and the rest of the world contributed 8%. Consolidated Q2 2026 gross profit grew 51% year-over-year to $7.2 million, with a gross margin of 30.4%. Non-GAAP operating expenses were $8.3 million, a 16% year-over-year increase (well below revenue growth, demonstrating operating leverage). The consolidated operating loss was $1.1 million, narrowed from $2.4 million in Q2 2025, and the net loss was $0.9 million, a 54% improvement year-over-year.

Risks & headwinds

- Forward-looking statements are subject to general market and operational risks that could cause actual results to differ materially from projected expectations, as disclosed in the company's SEC filings - Extended lead times for memory chips create supply chain and delivery risks, which the company is mitigating through proactive intentional inventory building - Rising memory and component costs create pressure on gross margins, requiring proactive sourcing, customer collaboration, and potential product spec adjustments to maintain margin targets - Rapid scaling of AI inference could require significant additional working capital, which the company has prepared for by maintaining a flexible balance sheet and an active shelf registration for financial agility

Analyst Q&A

• Q: What market verticals or use cases are driving the outperformance in the core business in the first half of 2026? Has per-unit pricing contributed to revenue upside? /

  A: Growth is broad-based across all core product lines, including FPGA, standard adapters, acceleration adapters, and edge systems. New product lines like switches, PQC, and AI inference do not yet contribute meaningful 2026 revenue, as they will ramp in future years. Gross margin is maintained through long-standing supplier relationships that enable favorable component pricing and availability, proactive customer collaboration on cost challenges, and intentional pre-purchasing of inventory enabled by the strong balance sheet.

• Q: With skyrocketing memory costs, are you or your customers moving to lower-memory product designs to offset costs? /

  A: The company works closely with each customer on an individual basis. Some customers have adjusted memory configurations (for example, reducing extra memory added when costs were lower) to hit target price points, while others have opted not to change specs. Silicom's ability to quickly customize products lets it smoothly transition customers to adjusted or new platforms when needed, with minimal disruption.

• Q: What is the approximate 2026 revenue contribution from newly secured 2026 design wins, particularly AI inference? /

  A: AI inference is expected to contribute $3 to $4 million in total 2026 revenue, with far higher contributions expected in 2027. Most 2026 design wins will not reach full run-rate revenue in 2026, with full contribution coming in 2027 or 2028 depending on the product timeline.

• Q: What is the purpose of the recently filed active shelf registration? Could you need to use all existing cash for working capital in the near term? /

  A: The shelf is standard corporate housekeeping to maintain maximum financial flexibility. It provides agility to support working capital needs if core business growth exceeds expectations, or if AI inference demand scales faster than anticipated, which would require significant inventory build. Management confirmed that rapid high-volume AI scaling is the most likely use for additional working capital.