Lattice (LSCC) Q2 2025 Earnings Transcript

Lattice Semiconductor reported Q2 revenue of $124 million, flat year over year, with non-GAAP gross margin at 69.3%, EBITDA margin at 34.1%, and EPS of $0.24, all at or above expectations. Management guided Q3 revenue to $128 million-$138 million, implying 7.2% sequential growth at the midpoint, with EPS of $0.26-$0.30 and gross margin of 69.5% +/- 1%, while highlighting strong AI/server demand, normalized communications and compute inventory, and improving industrial/auto trends. The company also repurchased $46 million of stock in the quarter and reiterated it is debt-free.

Analysis

LSCC is transitioning from a channel-destocking story into a mix-and-attach-rate story, and that matters more than the headline revenue beat. The important second-order effect is that AI infrastructure is creating a structurally denser bill-of-materials: as racks get more complex, Lattice’s low-power control/signal-management role becomes less cyclical and more embedded, which should support both ASP and design-win durability even if hyperscaler capex moderates. That shifts LSCC from a “recovering industrial semi” multiple to a quasi-razor/blade content play on AI infrastructure.

The market may still be underestimating how much of the near-term upside is self-help versus true end-demand. Inventory normalization in industrial/auto is effectively a future revenue release valve, but the more interesting leverage is margin: once that segment stops dragging and comms/compute stays tight, incremental gross profit should expand faster than revenue, especially if new products continue displacing legacy mix. That argues for outperformance into the next 2-3 quarters even without a heroic 2026 end-market forecast.

The main risk is that the current narrative is crowded: investors are already willing to pay for anything linked to AI capex, and LSCC’s small/midrange FPGA positioning is not a clean AI winner-take-all exposure. If hyperscaler ordering pauses, attach-rate gains can decelerate quickly because the company is still partly riding a capacity/content cycle rather than a purely secular software-like adoption curve. Tariff noise looks manageable for now, but the bigger latent risk is a 90-day lag between backlog strength and actual consumption if customers over-order into supply uncertainty.

Contrarianly, the consensus may be too focused on whether industrial/auto rebounds in 2026 and not enough on the fact that LSCC’s AI contribution is increasingly diversified across networking, board management, security, and edge inferencing. That breadth reduces single-customer dependence and makes the attach-rate thesis more durable than headline server exposure suggests. If management is right that new products keep compounding, the stock can re-rate before the full industrial recovery shows up in reported revenue.

Key Decisions for Investors

• Go long LSCC on a 3-6 month horizon; use pullbacks to add, targeting upside from margin expansion and AI attach-rate leverage. Risk/reward is favorable if revenue holds above the guided run-rate and gross margin stays near 69-70%.
• For traders, buy LSCC Jan-2026 calls or call spreads on weakness; the catalyst path is clear over the next 2-3 quarters as inventory normalizes and backlog converts. Limit downside with defined-risk structures given valuation sensitivity to any AI capex slowdown.
• Pair trade: long LSCC / short a more industrial-exposed semiconductor peer with weaker AI content and slower mix shift. The trade is designed to isolate the inventory-normalization tailwind and reduce macro beta.
• If you already own NVDA, consider LSCC as a complementary, higher-beta infrastructure-content satellite rather than a substitute. LSCC benefits from AI rack complexity without needing the same unit-growth risk as accelerators.
• Take profits on any sharp post-earnings move if LSCC rerates rapidly on AI enthusiasm; the setup is constructive, but the stock can overshoot fundamentals on crowded AI sentiment.