Lattice Semiconductor Corporation (NASDAQ: LSCC) — The Recovery Is Real; the Multiple Already Booked It Twice

⚡ Claude’s Take

This is the author’s own independent opinion and general information only — not investment advice, and not a recommendation to buy or sell any security. The detailed analysis that follows deliberately takes no position and carries no price target; the single subjective view is contained in this block alone.

Verdict: HOLD / own-for-quality / accumulate-only-on-a-real-pullback / not-a-short. A genuinely good fabless FPGA franchise, capably run, mid-cyclical-inflection — priced at the richest valuation in its own history (P/S in the 99.9th percentile of the last decade, ~37–40x trailing sales) that already discounts years of flawless execution plus a clean integration of a $1.65B levered acquisition the market is being handed for free. Directional zone: it is hard to argue the stock is anything but expensive within ~10–15% of $153. I’d frame fair value at roughly $95–120 (≈30–40x a normalized FY27 non-GAAP EPS of ~$2.75–3.25, or ~22–26x normalized EV/EBITDA), with a genuine accumulation zone only below ~$100, a trough/bear case toward $70–85, and a momentum bull case that can overshoot to $180+. Not a short — the business is inflecting, backlog runs into 2027, and the tape is a one-way street.

What the market is pricing correctly: this is a real business with a real (if narrow) moat — ~68–70% gross margins, ~30%+ through-cycle operating margins, asset-light, formerly net cash, ROIC in the high-20s%+ at mid-cycle, and an end-market that genuinely rotated from a cyclical industrial book into AI/data-center server attach in ~18 months (Comms & Computing +86% YoY in Q1’26). CEO Ford Tamer (ex-Inphi, ex-Broadcom) has executed cleanly — channel inventory down from ~6 months to under 2, EPS growing ~80% YoY, guidance beaten. What the market is mispricing: it has extrapolated the steepest part of a post-trough recovery into perpetuity at a peak multiple, and is treating an $1.65B (~8.25x sales) acquisition that flips the balance sheet from ~$92M net cash to ~$1B net debt — the largest deal in company history by 20x — as costless optionality. The framing is momentum/quality-at-the-wrong-price, not value: this is a high-beta (2.3), high-idiosyncratic-vol (41%), +200%-in-12-months name where the factor engine is “expensive semiconductor riding the AI tape,” and the LowVolatility loading is deeply negative. The recovery is genuine; the price has booked it about twice.

Conviction: medium. Flips bullish if AI-server FPGA attach proves a durable multi-year annuity (not a one-cycle capex spike) and AMI closes and integrates accretively — taking combined revenue convincingly through a >$1.5B run-rate at 35%+ margins, which would let earnings grow into the multiple. Flips bearish if the small-FPGA niche’s excess returns draw competition back faster than expected (a refocused standalone Altera under Silver Lake, China entrants on price, or hyperscaler SoCs absorbing the companion-chip socket), or if AMI integration disappoints and the leverage bites into a cyclical air-pocket. Tag: “Buy the franchise, not the print — and not at 40x sales.”

📈 Stock Price Action — Five-Year Event Map

LSCC has round-tripped and then some over five years: from ~$53 (mid-2021) up to a then-record ~$98 intraday in 2023, down to a five-year closing low of $37.84 (8 Apr 2025) in the depths of the inventory correction, and then a near-4x vertical move to an all-time high of $154.60 (3 Jun 2026), closing $153.72 on 18 Jun 2026 — within ~1% of the high. The 52-week range alone is $48.73–$154.60. The stock is at the very top of its own five-year cycle, on trough-to-recovering fundamentals.

The five-year price moves are FACT (daily price history); the attributed drivers are INTERPRETATION, cross-referenced to the revenue trajectory, the CEO transition, the channel-inventory normalization management has narrated, and the dated AMI announcement. No recommendation or target is implied here — the opportunity/mispricing judgment lives only in Claude’s Take above.

1. Executive Summary

Lattice Semiconductor is a fabless, low-power FPGA specialist — “the low power programmable leader” — headquartered in Hillsboro, Oregon (CIK 0000855658; ~1,174 employees; fiscal year ends the Saturday nearest 31 December). It designs small-to-mid-range field-programmable gate arrays and the software/IP that makes them usable, selling to OEMs across communications, computing, industrial, automotive and consumer end markets, ~84% through distributors and ~83% offshore (Asia 68%). It is not a diversified chip house; it is a focused programmable-logic franchise with a clean, high-margin model.

The investment debate is not about quality — it is about price. On quality, LSCC is genuinely good: ~68% GAAP / ~70% non-GAAP gross margins, ~30%+ through-cycle operating margins, asset-light (~$20M capex, <4% of sales), historically net cash, and mid-cycle ROIC in the high-20s%+. It holds a real — if narrow — competitive moat (toolchain lock-in, multi-year design-win/qualification cycles, low-power architecture IP, 20-year product longevity) that strengthened after AMD-Xilinx (2022) and Intel-Altera both vacated the low end. Under CEO Ford Tamer (ex-Inphi/Broadcom, since ~Oct 2024), execution has been crisp: the company worked channel inventory down from ~6 months to under 2, rotated its center of gravity from a collapsing industrial book into AI/data-center server attach, and is now compounding revenue +42% YoY and non-GAAP EPS ~+80% YoY (Q1’26), with backlog into 2027.

On price, the stock is at the richest valuation in its own ten-year history. Lattice trades at ~$153.72 (≈$21B market cap), or ~37–40x trailing sales and ~370x trailing EBITDA on still-trough financials — the 99.94th percentile of its own price-to-sales range and the 98.7th percentile on a composite of P/E, P/B and P/S. GAAP earnings are near-zero ($0.022 FY25 diluted EPS), distorted by a stock-based-comp charge that more than doubled to $115.6M (≈22% of revenue) — largely a one-time CEO inducement grant — so the headline 1,000x+ P/E is meaningless; the honest lenses are P/S and EV/sales, both at record highs. Layered on top is a $1.65B agreement to acquire AMI (firmware/BIOS/BMC/platform-security software; ~8.25x sales), funded with ~$1B of new debt that flips the balance sheet from ~$92M net cash to ~$1B net debt — strategically coherent (de-cyclicalizing the model and wrapping a software moat around the companion-chip hardware) but the largest deal in company history by more than 20x, struck at a peak multiple, part-funded with peak-multiple stock, and disciplined by a compensation plan that contains no return-on-capital metric whatsoever.

The embedded expectation at ~$21B EV is years of sustained 30%+ growth toward a $1.5–2B revenue base at 35%+ operating margins, plus flawless AMI integration — i.e., the AI-server attach must prove a durable annuity rather than a cyclical capex spike, the mid-range Avant push must scale against AMD/Altera, and the acquisition must compound rather than dilute. The bull case is real and live; the price simply leaves no margin for the many things that have to go right. This memo takes no position and sets no target; the analysis below frames the franchise, the moat, the financials, and the embedded expectations on the evidence.

2. Business Overview

What Lattice does. Lattice designs and sells low-power, small-to-mid-range FPGAs — reconfigurable logic chips — and the toolchains, IP cores and vertical software stacks that surround them. It self-describes consistently across filings as “the low power programmable leader” (FY25 10-K). The economic core is a single product category (programmable logic) sold to OEMs, supplemented by a small, lumpy IP-licensing / patent-monetization / design-services stream (HDMI/MHL royalties, IP-core licensing, occasional patent sales). It is a fabless model: chips are manufactured by third-party foundries and assembled/tested by partners.

The strategic frame. Under CEO Ford Tamer the company has crystallized a “companion chip / FPGA everywhere” thesis: position Lattice silicon not in competition with CPUs/GPUs/AI accelerators but as the small, always-on enabling chip beside them — hardware root-of-trust / secure boot (“first-on, last-off”), power sequencing, platform management, I/O aggregation, sensor bridging, and far-edge inference offload. Management frames roughly 3 FPGAs per AI server, with server applications rising to ~38% of 2026 revenue and AI specifically to ~25% (from the mid-teens in 2024) — management commentary, treated as hypothesis, but corroborated by the segment trajectory below.

Product families (platform-based since 2019). One foundational silicon platform spawns multiple families:

• Nexus (small-FPGA platform, on Samsung 28nm FD-SOI) — the volume engine: CrossLink-NX (video/MIPI connectivity), Certus/CertusPro-NX (general purpose), MachXO5-NX/MachXO5D-NX (control & hardware root-of-trust; the TDQ variant claims industry-first post-quantum / CNSA 2.0 support).
• Nexus 2 (2024, TSMC 16nm) — next-gen small-FPGA platform (Certus-N2).
• Avant (mid-range platform, 2022, TSMC 16nm FinFET) — the up-market push into the 150K–600K+ logic-cell range (Avant-E/G/X), the segment AMD/Altera deprioritized at the bottom but still hold at the top.
• Legacy long-life families (MachXO/2/3/4, CrossLink, ECP, iCE40 UltraPlus) — explicitly marketed with 20+ year product longevity, a genuine sticky annuity in industrial/automotive.
• Software/solution stacks: Radiant & Diamond (design tools), Propel (embedded), plus sensAI/Sentry/mVision/Automate/Drive vertical stacks.

Revenue by end market ($000s; FY25 10-K). The mix shift is the whole story:

At the FY23 peak, Industrial & Automotive was 59% of revenue; by FY25 it had fallen ~55% in dollars and to 37% of mix as the post-COVID industrial inventory glut purged. Meanwhile Comms & Computing (data-center servers — general-purpose and AI — plus wireline) grew +28% in FY25 and is now the majority (56%), accelerating to +86% YoY in Q1’26. Lattice’s center of gravity flipped from cyclical broad-market industrial to AI/data-center server attach in roughly 18 months. As of Q1’26 the company recast its reporting into two segments — Compute & Communications (62% of Q1’26 revenue) and Industrial & Embedded (now folding in consumer).

Geography & channel. Asia is 67.6% of FY25 revenue (rising), Americas 19.6%, Europe 12.8%; foreign sales ~83%. ~84% of revenue flows through distributors, and concentration is high and rising: two distributors = ~69% of total revenue (Distributor B alone jumped to 38% from 21% two years earlier, and was 62% of year-end receivables). This is sell-in accounting one step removed from end-demand — a structural quality flag to monitor.

Recurring vs. non-recurring. There is no contractual/subscription revenue. The functional annuity is (a) long-life industrial/auto design-win sockets that re-order for years, and (b) HDMI/MHL royalties. IP/patent-sale revenue is explicitly non-recurring and lumpy. The new AI/server revenue is design-win-driven (durable per socket) but young enough that re-order durability is unproven.

Verdict: A focused, fabless, high-gross-margin programmable-logic specialist with genuine product breadth inside its niche. Two structural flags — ~69% distributor concentration on sell-in accounting, and a forward thesis now concentrated in a single, currently-euphoric AI/data-center driver after a violent rotation out of industrial. The business is real and differentiated; the open question is whether the AI driver is a durable annuity or a cyclical spike.

3. Industry Dynamics

Structure and size. FPGAs are a high-value, high-margin, oligopolistic corner of semiconductors — a TAM on the order of $10–13B (third-party estimates; Lattice does not publish one). The structure matters more than the size: FPGAs sustain 60–70%+ gross margins because the moat is the software/IP ecosystem and multi-year design-win lock-in, not the silicon. Historically the market was a near-duopoly (Xilinx + Altera ≈ ~85% combined) with a long specialist tail.

The competitive set has been reshaped by consolidation and re-spin — the single most important industry fact for this thesis:

• AMD acquired Xilinx (closed Feb 2022, ~$49B). Inside AMD, the franchise is steered toward large, high-end data-center adaptive-SoC FPGAs (Versal), structurally deprioritizing the small/cost-sensitive low end — the vacuum Lattice colonized.
• Altera spun back out of Intel. Intel bought Altera (2015, ~$16.7B), underperformed, and carved it out: Silver Lake agreed to a 51% majority stake (April 2025, valuing Altera ~$8.75B), making Altera a standalone, PE-backed company again. This is double-edged for LSCC: under Intel, Altera was a neglected asset that ceded low-power share; refocused and re-capitalized, it could re-engage the mid-range (where Avant is newest) and low end. This is the key forward swing factor.
• Microchip (PolarFire/IGLOO/SmartFusion) — the most direct overlap in low-power, flash-based, security-oriented small/mid FPGAs; pressure muted recently only by Microchip’s own deep inventory downcycle.
• China / emerging entrants — GOWIN (aggressively priced small FPGAs, the most direct iCE40/MachXO price threat in cost-sensitive Asian designs), Efinix (well-funded edge-AI), Achronix (high-end, struggling), QuickLogic (tiny eFPGA). The real long-run margin threat is not engineering but erosion of the price umbrella in commoditizing China sockets — exactly where LSCC’s Asia ship-to concentration is highest.

Barriers to entry (Greenwald). Three reinforcing barriers: (1) software/toolchain ecosystem — an FPGA is useless without its synthesis/place-and-route/IP toolchain, and customer investment in learning it and building on-fabric IP is non-portable (the genuine switching cost); (2) long design-win + qualification cycles — 12–24+ months to win a socket, multi-year for auto (AEC-Q100) and defense, after which the part ships for the product’s 5–20-year life; (3) architecture/IP + multi-node manufacturing know-how accumulated over decades. These barriers are real but graduated — strongest in auto/industrial/defense, weakest in fast-cycle consumer and price-sensitive China designs.

Pricing power. Within the niche, Lattice has shown real pricing power: gross margin expanded from the low-60s% (pre-2019) to ~68–70% as Nexus/Avant displaced legacy parts and the low-end vacuum reduced discounting. But this is contingent on the duopoly’s continued neglect of the low end and on China entrants not yet being qualified into Western OEM sockets — both reversible.

Demand drivers. Edge/far-edge AI (sensor bridging/aggregation/offload), data-center server attach (secure boot, power sequencing, platform management, I/O — the live driver), post-quantum cryptography / hardware security (programmability beats fixed-function ASICs for evolving crypto), plus longer-dated robotics, ADAS, electrification and 5G/ORAN optionality.

Capital-cycle read (Marathon). The industry sits at a bifurcated point: AI/data-center is in a euphoric, capital-attracting up-leg (the classic warning sign), while broad industrial/auto is emerging from a deep 2023–25 inventory-correction down-leg (the part Marathon likes — capital starved, capacity rationalized). LSCC straddles both — its industrial book is mid-recovery from a –55% purge (favorable supply-side setup), while its new AI/server revenue plugs directly into the euphoric leg (unfavorable for mean reversion). Critically, capital is re-entering the FPGA niche itself — the Altera carve-out and well-funded China/Efinix entrants are the textbook “high returns attract capital” signal.

Verdict: Structurally good — among the better corners of semiconductors — but with a deteriorating supply-side setup at the margin. A high-margin oligopoly defended by software lock-in and long design cycles, where post-Xilinx/Altera consolidation genuinely opened the low end. But the structure is not static: a refocused standalone Altera, China entrants eroding the price umbrella, and AI-driven capital inflows all signal that the niche’s excess returns are beginning to attract the capital that erodes them. Good industry, late-ish cycle.

4. Competitive Position

Naming the moat (Greenwald). Lattice’s advantage is best classified as customer captivity + intangibles (toolchain lock-in, design-win lock-in, low-power architecture IP), reinforced by a focused-scale/cost advantage within the low-power niche — not a network effect, and not a broad economies-of-scale moat (it is sub-scale vs. AMD/Altera in absolute terms). The Greenwald tests — stable/rising share, ROIC > cost of capital, and financial deterioration if the moat vanished — pass for the core niche: ~68–70% gross margins, high-20s%+ mid-cycle ROIC, and demonstrated low-power small-FPGA share gains. The moat is real but narrow, and the bull claims need testing one by one.

Bull claim (a): “Low-power leadership after Xilinx & Altera left the low end.” Largely true, the strongest leg. AMD-Xilinx and Intel-Altera genuinely steered toward large high-end FPGAs; Lattice built Nexus into that vacuum and grew share, corroborated by margin expansion. The risk: this rests on a competitive choice by larger rivals, not an insurmountable barrier. Standalone Altera (Silver Lake, 2025) is the live threat — it retains the Quartus toolchain, a deep IP library, and incumbent sockets, and could re-engage the mid-range where Avant (launched 2022) is LSCC’s most exposed flank — attacking up-market into AMD-Versal/Altera-Agilex strength, unproven at scale.

Bull claim (b): “Switching costs from toolchains + multi-year design-win lock-in.” True, the most durable leg. Toolchain learning curves and on-fabric IP are genuinely non-portable; long design + qualification cycles + 20-year longevity make sockets sticky for years. This is the real customer-captivity moat that produces the margins. The caveat: switching costs lock in existing sockets but do not win new designs — every new design is re-competed. So the moat protects the installed industrial/auto annuity but must be re-earned on the AI/server wave now driving growth. The lock-in is asymmetric.

Bull claim ©: “Companion-chip attach + security/root-of-trust.” Real today, the most strategically fragile leg. The 3-FPGAs-per-server thesis is shipping and is the engine of +86% Q1’26 Comms & Computing growth; the PQC/root-of-trust angle is credible. The core bear risk: integration absorbs the socket. Companion-chip “glue logic” is exactly what has historically been absorbed into larger SoCs/ASICs once volume justifies it (the recurring Moore’s-Law threat to FPGAs). If hyperscaler custom AI-server SoCs or BMC/platform-controller vendors integrate secure-boot/power-sequencing/management on-die, per-server FPGA attach could compress even as server volumes rise. Tellingly, the $1.65B AMI acquisition reads partly as a defensive move to own the platform-management/security software stack and lash the FPGA to a sticky software layer — i.e., management itself appears to judge that the hardware-only companion-chip socket is not sufficiently defensible alone.

Direct competitive comparison:

Verdict: A real but NARROW moat — durable in the defensible low-power/security niche, weaker on the contested growth frontier. Lattice has a genuine customer-captivity + intangibles moat (toolchain lock-in, long design/qual cycles, low-power IP, 20-year longevity), validated by ~68–70% gross margins and post-consolidation share gains. But it is graduated: rock-solid on legacy industrial/auto/defense sockets, materially thinner on the two things now driving the stock — the up-market Avant push and the AI-server companion-chip attach. The Altera re-spin and China entrants are the capital-cycle signals that the niche’s excess returns are beginning to attract competition. Own the moat; don’t mistake the niche for an unassailable franchise. The two falsification tests: AI-attach durability and whether standalone Altera re-engages.

5. Growth History and Forward Opportunities

History — a textbook semiconductor cycle, with a structural mix-shift layered on top. Revenue ran $408M (2020) → $515M (2021) → $660M (2022) → $737M (2023 peak) → $509M (2024, –31%) → $523M (2025, +2.7%). The 2022–23 surge was the post-COVID up-cycle (scarcity, pricing power, Nexus ramp); the 2024 collapse was a brutal industrial/auto inventory correction (Industrial & Auto fell ~55% peak-to-trough). What makes this cycle different from a plain bounce is the composition of the recovery: the industrial book is only mid-recovery, but Comms & Computing (data-center/AI server attach) grew through the downturn (+28% in FY25) and has become the majority of revenue, accelerating to +86% YoY in Q1’26. Total Q1’26 revenue was $170.9M, +42% YoY and +17% QoQ, with Q2’26 guided to ~$185M midpoint (≈+50% YoY). This is a genuine inflection, not stabilization.

The growth is currently high-quality on the metrics that matter: it is volume-and-content-led (rising FPGAs-per-server and rising ASP per unit, per management), it is dropping through at high incremental margins (Q1’26 non-GAAP operating margin 34.4%, EBITDA margin 39.6%, with EPS +80% YoY — operating leverage is real and visible), and it is backed by bookings extending into 2027 and channel inventory normalized to under 2 months (from ~6 months when Tamer arrived). The caveat is concentration and durability: the growth lives in the least-moated, most cyclical part of the franchise (AI/data-center capex), and AI-attach re-order durability across a full cycle is unproven.

Forward opportunities. (1) Data-center server attach — the live engine: rising server units, rising AI mix, rising FPGA content per box for secure boot, power/cooling rack management, I/O aggregation, and “rack boot” use cases; the rack disaggregation trend (separate compute/networking/power/cooling racks) management describes would lift comms and power/cooling content. (2) Avant mid-range ramp — the up-market TAM expansion, but the most contested. (3) Edge/physical AI, robotics, ADAS, PQC/security — longer-dated. (4) AMI — management frames the acquisition as roughly doubling the serviceable market from ~$6B to ~$12B over 3–4 years and accelerating combined growth (and combined revenue through a >$1B run-rate exiting 2026). That SAM-doubling is a management projection, not a fact, and depends on cross-selling software + FPGA “solutions” that do not yet exist at scale.

Verdict: High-quality growth today — volume + content + ASP, dropping through at high incremental margins — but concentrated in a single euphoric driver and partly cyclical. The inflection is real and the operating leverage is genuine; the risk is extrapolating the steepest part of a post-trough, AI-capex-fueled recovery as if it were a secular straight line.

6. Financial Quality

Margins & operating leverage. Gross margin is structurally high and stable — ~68% GAAP / ~70% non-GAAP — reflecting the value-based pricing of differentiated low-power parts. The operating-leverage story is the headline: GAAP operating margin ran 29% (2023) → 11.9% (2024) → 2.9% (2025), but that trough is heavily distorted by stock-based comp (below); on a non-GAAP basis the model snapped back to 34.4% operating / 39.6% EBITDA margin in Q1’26, with management guiding sustained earnings growth well ahead of revenue. Through-cycle, this is a 30%+ operating-margin, ~68% gross-margin franchise.

The quality-of-earnings crux — SBC, and why GAAP EPS is meaningless this year. FY25 GAAP diluted EPS was $0.022 (net income just $3.08M) — hence the absurd ~1,000x trailing P/E. This is not a sign of a broken business; it is the collision of (a) trough revenue with (b) a stock-based-compensation charge that more than doubled: SBC ran $70.2M (2023) → $53.0M (2024) → $115.6M (2025, ≈22% of revenue). The spike is largely a one-time event: CEO Ford Tamer’s 2024 inducement/hiring equity grant (2024 proxy summary-comp total ~$69.15M, including a sign-on package partly replacing forfeited prior-employer equity), front-loaded and amortizing under ASC 718. Tamer received no new equity in 2025 (2025 SCT ~$1.73M), so SBC at ~22% of revenue is a transition artifact, not run-rate — it should normalize toward ~10–13% of a recovering revenue base as the grant vests. The honest readings of this year’s earnings are therefore (i) non-GAAP EPS (Q1’26 $0.41, +80% YoY; Q2’26 guide ~$0.44) and (ii) cash flow, not GAAP net income.

Cash flow. FY25 operating cash flow was $175.1M and capex only $19.8M (asset-light, <4% of sales), for free cash flow of ~$155M — i.e., FCF was ~50x GAAP net income, the entire gap being non-cash SBC and D&A. The intellectually honest “owner” figure deducts SBC as the real economic cost it is: owner-FCF ≈ FCF $155M − SBC $115.6M ≈ $40M in FY25 — but because the SBC is inflated by the one-time grant, normalized owner-FCF on a recovering revenue base is materially higher (a run-rate SBC of ~$60–70M against rising OCF would put normalized owner-FCF comfortably above $150M as revenue scales). The point for valuation: do not pay the headline FCF multiple as if SBC were free, but also do not annualize the FY25 SBC drag as permanent.

Returns on capital. Through the cycle this is a high-return business: ROIC ~17% (2021) → ~30% (2022) → high-20s/30s% (2023) → collapsing to ~0.5% (2025 trough) purely on the earnings trough. ROA was 31.6% in 2023. With ~68% gross margins, minimal capex, and (until AMI) net cash, mid-cycle ROIC comfortably exceeds the cost of capital — the financial signature of a real moat. The 2025 prints are trough artifacts, not the steady state.

Balance sheet. As of FY25: cash $133.9M, debt only $42M (capital leases) → net cash ~$92M. Equity $714M; current ratio 3.1x; inventory worked down to $89M from $103M. Goodwill $315M + intangibles ~$4M means tangible book is ~$395M (the high P/B is partly a goodwill/asset-light artifact). This pristine balance sheet is about to change: the AMI deal adds ~$1B of new debt, flipping the company to ~$1B net debt (~2–2.5x trailing combined EBITDA, de-levering quickly given ~40% combined FCF margin per management). Manageable, but a genuine change in the risk profile.

Verdict: Economics clearly improve with scale, and the model is genuinely high-quality through the cycle — but FY25 GAAP earnings are uninformative. Judge this business on non-GAAP earnings power and cash flow (both strong and inflecting), not on the SBC-and-trough-distorted GAAP line. The one real degradation is the balance sheet: a formerly net-cash franchise is taking on ~$1B of debt to buy growth.

7. Capital Allocation

The recurring decisions have been competent-to-good; the franchise-defining decision is a big, leveraged, peak-multiple acquisition the comp plan does nothing to discipline.

Buybacks — well-timed. In FY25 Lattice repurchased 1,763,053 shares for $100.0M at an average ~$56.72 — against a current price near $153, a genuinely value-accretive use of cash, not just dilution-mopping. Net share count actually fell (137.70M → 136.77M) despite the SBC spike. Q1’26 added $15M of buyback; a new $250M authorization was approved in December 2025. No dividend has ever been paid (appropriate for a growth-stage chip company).

R&D — productive and counter-cyclically funded. R&D was $188.0M in FY25 (~35.9% of trough revenue, +18% YoY) — held up and grown through the downturn, and it visibly worked: it funded the Nexus 2/Avant roadmap and the server/AI mix shift that is now driving growth. The 35.9% ratio is a trough-denominator artifact (~22% at the FY23 peak).

Compensation & incentive alignment — no return-on-capital governor. From the 2026 proxy: the short-term incentive (2025 Corporate Incentive Plan) is three equally weighted metrics, 200% cap — (i) non-GAAP operating income, (ii) GAAP revenue, (iii) MBOs (note non-GAAP op income excludes SBC). The long-term plan is >57% of NEO equity in relative-TSR PSUs (vs. Russell 2000/3000) plus revenue-growth PSUs (FY26–29, threshold 5% / target 10% / max ≥25%, gated on beating the Gartner non-memory-semiconductor benchmark) plus a “Go for Gold” revenue-growth program. No ROIC / ROE / return-on-capital metric appears anywhere. Pay is 100% levered to growth, SBC-adjusted profit, and relative TSR — a structure that rewards exactly the kind of large, growth-buying, leverage-adding acquisition the company just announced, and does nothing to penalize value-destructive empire-building. Governance hygiene is otherwise solid (clawback, anti-hedge/pledge, no repricing, CEO ownership requirement raised to 5x salary, independent comp consultant).

Insider behavior — a notable absence of conviction. A sweep of the trailing-5-year Form 4 corpus (≈694 filings since Jan-2021) found zero code-P open-market purchases — every transaction is a sale (frequently 10b5-1), tax-withholding, or grant/exercise. No insider bought a single share in the open market even at the April-2025 trough of $37.84. Insider ownership is thin: CEO <1%, all insiders ~0.62%. This is not damning on its own (common for professionally-managed mid-cap tech), but it means there is no insider “skin-in-the-game” signal corroborating the stock’s quadrupling.

The AMI acquisition — strategically coherent, financially aggressive, at a peak. $1.65B ($1.0B cash + $650M equity ≈ 5.4M shares, ~4% dilution) for a ~$200M-revenue firmware/BIOS/BMC/platform-security software business — ≈8.25x sales for a high-teens grower. Strategically it is defensible: AMI’s software is adjacent to the FPGA companion-chip socket (server platform management/security), it is asset-light and high-margin (GM above Lattice’s, EBITDA margin similar), it de-cyclicalizes the model with recurring software, and management projects immediate non-GAAP accretion to GM/FCF/EPS without relying on synergies. But the financing and timing carry real risk: ~$1B of new debt flips the balance sheet from net cash to ~$1B net debt; the equity slug is paid in peak-multiple stock; and this is the largest deal in company history by more than 20x (the only prior acquisition of note was Mirametrix at ~$68M in 2021) — i.e., an unproven integration muscle being exercised at maximum scale. Through the Marathon lens, it is the textbook late-cycle move: using a richly-valued currency to buy growth at a rich multiple, at a moment of peak optimism.

Verdict: Above-average on the recurring decisions (well-timed buybacks, productive counter-cyclical R&D, no dividend trap), but the thesis is now hostage to one large, leveraged, peak-multiple acquisition that the compensation plan — which lacks any return-on-capital metric — actively incentivizes rather than restrains. A capable operator of cash being levered into an unproven, full-priced M&A bet. Watch the integration and the de-levering closely.

8. Changes and Headwinds — Last Two Years

Strategic/leadership. The defining change is the arrival of CEO Ford Tamer (~Oct 2024), ex-Inphi CEO (which he built into a ~$10B optical-interconnect leader sold to Marvell) and ex-Broadcom. He brought a sharper “companion chip / system-solutions” strategy, normalized channel inventory (from ~6 months to under 2), recast reporting into Compute & Communications vs. Industrial & Embedded, and — most consequentially — pivoted from organic-only to a major M&A footing with the AMI acquisition (announced 4 May 2026, $1.65B, closing ~Q3’26). CFO is Lorenzo Flores.

Operational/financial. The company round-tripped a full cycle: 2023 peak → 2024 inventory-correction trough → 2025–26 inflection (Q1’26 revenue +42% YoY, EPS +80% YoY, backlog into 2027). The end-market mix flipped from industrial-led to AI/data-center-led. SBC more than doubled (one-time CEO grant). The balance sheet is about to lever up ~$1B for AMI. A new $250M buyback authorization was set (Dec-2025).

Headwinds / watch-items. (1) Valuation — at the richest multiple in company history, the bar for incremental performance is extreme. (2) Competitive re-entry — standalone Altera (Silver Lake), China entrants (GOWIN/Efinix), and Microchip emerging from its own downcycle. (3) AI-attach durability — whether server-attach revenue is a multi-year annuity or a capex spike. (4) AMI integration + leverage — first large deal, net-debt risk into any cyclical air-pocket. (5) Distributor concentration — ~69% through two names on sell-in accounting. (6) Geopolitics/supply — 83% offshore revenue, Asia 68%, single-foundry node dependence (Samsung 28nm FD-SOI for the volume Nexus line); management flags back-end (assembly/test) supply as the current constraint in the up-cycle. (7) Supply-cost inflation — management expects gross-margin cost pressure to build in H2’26.

Verdict: The operational changes are thesis-strengthening (capable CEO, clean inflection, real growth); the financial/strategic changes (peak multiple, levering up for a large first-time acquisition) raise the risk profile. Net, the franchise is in better operational shape than two years ago and in a more precarious valuation-and-balance-sheet position.

9. Risk Analysis (Risk Matrix)

The dominant risk is simply valuation × any disappointment: at a record multiple on still-recovering financials, the asymmetry is negatively skewed — modest good news is priced, while any stumble in the AI driver, the competitive set, or the AMI integration would compress a 40x-sales multiple violently. Catastrophic permanent loss is unlikely (real franchise, manageable leverage, net positive equity); a 40–60% drawdown on a multiple reset + cyclical air-pocket is well within the historical range (the stock fell ~60% in 2023–25 and has a –85% lifetime max drawdown).

10. Valuation Discussion (Embedded Expectations)

No price target and no recommendation in this section — embedded-expectations and scenario framing only.

Where the multiple sits. At ~$153.72 (~$21B market cap; pro-forma ~$22.9B EV after AMI net debt), LSCC trades at the richest valuation in its own ten-year history: AZI’s own-history percentiles put the composite at the 98.7th, P/S at the 99.94th (~37x trailing sales), and P/B at the 99.4th. The GAAP P/E (~1,000x+, 96.8th percentile) is uninformative — it reflects the SBC-and-trough-distorted $0.022 EPS, not earning power, and should be ignored. The honest lenses are price/sales and EV/sales, both at record highs, and forward non-GAAP P/E. On forward earnings: annualizing the Q1–Q2’26 trajectory implies ~$1.85–2.00 of non-GAAP EPS in 2026 (≈80x forward) and perhaps ~$2.75–3.25 combined in 2027 (≈48–56x). On sales, ~$22.9B EV against ~$1.0–1.2B of 2027E combined revenue is ~19–23x forward sales — still extreme for a hardware-led business, even a 70%-gross-margin one.

Comp context (own-history, not cross-sectional). Even versus its own peak-cycle multiples, today is richer: in the 2023 boom LSCC topped out near ~18x EV/sales and ~36x trailing EV/EBITDA; today’s trailing EV/sales (~40x) is more than double that, on lower absolute revenue. Among factor-similar semis (Microchip, ON, NXP, ADI, Entegris, Diodes — all ~0.9 factor-correlation), LSCC screens as one of the highest price-to-sales names, justified bulls would argue by its growth rate and gross margin — but the gap is wide.

Embedded-expectations reverse read. To justify ~$21–23B of enterprise value at a “normal” mature multiple (say ~25–30x non-GAAP earnings or ~18–20x EV/EBITDA at maturity), Lattice would need to grow into roughly $1.5–2.0B of revenue at 35%+ operating margins within ~4–5 years — i.e., revenue must roughly triple from the FY25 base, the AI-server attach must prove a durable multi-year annuity (not a one-cycle spike), the Avant mid-range push must scale against AMD/Altera, and AMI must integrate and compound rather than dilute. None of those is implausible; all of them happening cleanly, in sequence, is what the current price already underwrites. The market is paying for the bull case as the base case.

Scenario framing (illustrative, not targets):

• Bear: AI attach normalizes as a capex cycle rolls over and/or Altera re-engages; revenue stalls in the $700–900M range at compressed margins; AMI integration disappoints into ~$1B of debt. A multiple reset toward ~12–15x EV/sales (still rich) on lower numbers implies a large drawdown.
• Base: combined revenue compounds to ~$1.3–1.6B by 2028 at ~33–35% operating margins; non-GAAP EPS ~$3.50–4.50; the multiple normalizes from record highs toward a (still premium) ~30–35x — total return roughly tracks earnings growth net of de-rating, i.e., modest from here.
• Bull: AI attach is a secular annuity, content-per-server keeps rising, Avant scales, AMI doubles the SAM and combined growth holds 25–30%+; revenue toward $2B+ at 38%+ margins; the multiple stays elevated on momentum — meaningful further upside, but predicated on near-flawless execution.

The honest summary: the franchise quality is not in question; the price is. At a record multiple on recovering-but-still-trough financials, with a large levered acquisition layered on, the embedded expectation is for the best case to materialize. That is a poor risk/reward symmetry even for a good business.

11. Variant Perception

Consensus belief. LSCC is a high-quality, secular-growth FPGA winner — the low-power leader that took share when Xilinx/Altera left the low end — now riding an AI/data-center server-attach supercycle with a proven new CEO, expanding margins, ~80% EPS growth, and a SAM-doubling AMI acquisition. Sell-side and the tape treat it as a “own-the-quality-and-the-AI-theme” compounder; the +200% 12-month move and 99.9th-percentile multiple reflect that consensus.

Strongest bull case. The end-market rotation is structural, not cyclical: FPGAs are uniquely suited to AI-server companion functions (security/PQC, power, management, I/O) that scale with server units and content-per-box; the moat (toolchain lock-in, long design wins) protects the annuity; the new CEO is a proven value-creator (Inphi); and AMI transforms the model into a higher-margin, recurring, software-plus-hardware “secure management platform” with double the SAM. Earnings can grow into the multiple over a few years, making today’s price look reasonable in hindsight.

Strongest bear case. This is a cyclical semiconductor at a cyclical peak, dressed as a secular compounder, at its richest-ever multiple. The growth lives in the least-moated, most cyclical part of the franchise (AI capex); the niche’s excess returns are already attracting capital back (standalone Altera, China entrants); the companion-chip socket is structurally exposed to SoC integration (which the AMI deal implicitly concedes); GAAP earnings are near-zero; the balance sheet is being levered ~$1B for a first-ever large acquisition at ~8x sales; insiders own ~0.6% and have bought nothing; and the comp plan has no return-on-capital discipline. A multiple reset on any growth wobble is the base-rate outcome.

The 3–5 assumptions that matter most:

1. Is AI-server FPGA attach a durable annuity or a capex spike? (The whole growth thesis.)
2. Does the low-power niche’s moat hold against re-entering capital (standalone Altera, China, SoC integration)?
3. Will AMI integrate accretively and de-lever as promised — or become a distraction/value-leak?
4. Can margins hold ~35%+ through supply-cost inflation and the eventual industrial up-cycle normalization?
5. Does the record multiple compress, and how fast — the dominant driver of forward returns regardless of fundamentals.

Falsification evidence. Bull case breaks if: server-attach revenue decelerates sharply on a data-center capex pause, OR Altera/China visibly take a flagship socket, OR AMI integration produces write-downs/guide-downs. Bear case breaks if: combined revenue compounds 25%+ for multiple years at 35%+ margins with AI-attach re-orders proving sticky, AND AMI accretes and de-levers on schedule — in which case earnings genuinely grow into the price.

Factor-positioning read (input, not a call). A quantitative factor-model analysis confirms the framing: market beta ~1.6 (all-factor model) / ~2.3 raw, R² ~0.61, idiosyncratic volatility ~41% (high stock-specific risk), with a strongly negative LowVolatility loading (–0.87, i.e., a high-vol name) and positive Semiconductor (+1.13) and Quality (+0.28) loadings — but, tellingly, no active standalone Momentum loading (the move is absorbed by the market/semi/vol factors). The risk-adjusted record is extreme and recent: +201% trailing 12 months, a 12-month relative-strength reading in the top decile, and a 3-month annualized return that de-annualizes to ~+73% — i.e., a near-vertical, one-way tape. Lifetime stats remind that this is a violent name (17.6% annualized lifetime return, but a –85% lifetime max drawdown and –61% drawdowns within the last decade). The positioning read says: this is a crowded, high-beta, high-idio-vol semiconductor riding the AI tape near the top of its range — a momentum-quality name, not a value or contrarian setup. Consensus is with the stock, which is exactly when crowded longs are most offsides if the AI-attach narrative slips.

12. Fact vs. Interpretation Table

13. Open Questions

1. AI-attach durability: Are data-center server FPGA sockets multi-year re-ordering annuities, or will they fade with the AI-capex cycle? (Unanswerable until a full cycle is observed.)
2. AMI economics: What are AMI’s actual revenue growth, gross margin, EBITDA, and recurring-vs-license mix? (Management deferred full detail to post-close, ~Q3’26.)
3. AMI price: Is ~8.25x sales for a high-teens-growth software asset disciplined or rich? What is the implied return on the ~$1.65B (with $1B debt) over a reasonable horizon?
4. Avant traction: Is the mid-range push winning real sockets at scale against AMD-Versal / Altera-Agilex, or is it still early/sub-scale?
5. Altera re-engagement: Will a refocused, PE-backed Altera actually re-enter the low/mid-range, and on what timeline?
6. Normalized SBC run-rate: Where does SBC settle as the CEO grant vests — 10%, 12%, 14% of revenue? (Drives owner-FCF.)
7. Distributor B: Why is one distributor’s share climbing so fast (21%→38% in two years), and how much is sell-in vs. true sell-through?
8. Margin sustainability: Can ~70% gross / ~35% operating margins hold through H2’26 supply-cost inflation and the eventual industrial up-cycle mix normalization?

14. What Must Be True

Bull case — what must be true:

• AI/data-center server FPGA attach is a durable, multi-year, re-ordering annuity with rising content-per-server and ASP, not a one-cycle capex spike.
• The low-power/security niche moat holds — Altera does not re-engage materially, China entrants stay out of Western OEM sockets, and SoC integration does not absorb the companion-chip socket.
• AMI integrates accretively and de-levers (~40% combined FCF margin), proving the model can be transformed into higher-margin software-plus-hardware without value leakage.
• Combined revenue compounds toward $1.5–2B at 35%+ operating margins, letting non-GAAP earnings grow into the multiple.
• Falsification test: two consecutive quarters of Comms & Computing deceleration to flat/negative QoQ on a data-center capex pause, OR a public flagship socket loss to Altera/China, OR an AMI-related write-down/guide-down — any one breaks the bull case.

Bear case — what must be true:

• The AI-attach revenue is substantially cyclical, peaking with the current data-center capex wave.
• The niche’s excess returns draw competition back (standalone Altera, China price erosion, hyperscaler SoC integration), compressing share and/or margins.
• The record multiple compresses materially from 99.9th-percentile levels as growth normalizes, with the ~$1B AMI debt amplifying the downside in any cyclical air-pocket.
• Falsification test: combined revenue grows 25%+ for multiple consecutive years at 35%+ operating margins with AI-attach re-orders demonstrably sticky, AND AMI accretes and de-levers on schedule — that breaks the bear case and validates the secular-compounder framing.

15. Source Appendix

See Appendix B for the full source list with URLs and access dates. Primary sources: LSCC FY2025 10-K (filed 2026-02-13, CIK 0000855658), FY2023 10-K (2024-02-16), DEF 14A proxies (2026-03-18, 2025-03-20), Q1 2026 earnings call transcript (4 May 2026), and the AMI acquisition announcement (4 May 2026). Quantitative data: SEC EDGAR XBRL, public financial-data aggregators (statements, ratios, enterprise value, valuation multiples), own-history valuation percentiles, daily price history, and a quantitative factor model. All management commentary is treated as a hypothesis and validated against filings and financials.

APPENDIX A — Standard Diligence Questionnaire

Supplemental to the analysis above. Fact/Interpretation/Assumption labels applied where it matters.

General

What thoughtful questions have other investors asked about this company? The recurring buy-side questions cluster on: (1) AI-server FPGA attach durability — is it a secular annuity or a capex spike, and what is the real content-per-server / ASP trajectory? (2) AMI economics and price — what are AMI’s actual margins/growth and is ~8.25x sales disciplined? (3) Altera re-engagement — does a Silver Lake-backed standalone Altera re-enter the low/mid-range? (4) Margin sustainability — can ~70% gross / ~35% operating margins hold through supply-cost inflation and industrial-mix normalization? (5) Valuation — how can a 40x-sales multiple on trough financials be justified? (6) SBC normalization — where does stock-comp settle as the CEO inducement grant vests?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? Interpretation: GAAP earnings are at a cyclical low (FY25 net income $3.08M / $0.022 EPS), distorted by both trough revenue and a one-time SBC spike. Non-GAAP/cash earnings are inflecting off a trough — recovering fast (EPS +80% YoY in Q1’26) but not yet at a cyclical high; the prior peak was FY23 ($259M GAAP NI). Revenue is mid-recovery ($523M vs. $737M peak).

Driven by external environment or internal actions? Both. The downturn was an industry-wide inventory correction (external); the recovery is part cyclical (industrial restock) and part internal/structural (the deliberate rotation into data-center/AI server attach and channel-inventory discipline under the new CEO).

How stable are revenues? Cyclical — this is a semiconductor with a full peak-to-trough-to-recovery on display (−31% in 2024). The industrial/auto book is the cyclical core; the long-life design-win sockets and royalty streams provide a partial stable base; the new AI book is high-growth but unproven across a cycle.

Outlook for products/services? Strong near-term (backlog into 2027, Q2’26 guide ~+50% YoY); the secular question is AI-attach durability and Avant mid-range traction.

How big will this market be? FPGA TAM ~$10–13B (interpretation, third-party estimates); management frames its serviceable market doubling from ~$6B to ~$12B over 3–4 years with AMI (management guidance — hypothesis). Growing, global (83% foreign revenue).

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Interpretation: more competitive at the margin — capital is re-entering the FPGA niche (standalone Altera under Silver Lake, well-funded China entrants GOWIN/Efinix, Microchip emerging from its downcycle). The post-Xilinx/Altera vacuum that helped LSCC is beginning to close.

How profitable is the business (ROIC, ROE)? High through-cycle: ROIC ~17%→30%→high-20s% (2021–23), collapsing to ~0.5% at the 2025 trough purely on the earnings trough; ROA 31.6% in 2023. ~68% gross margin, ~30%+ through-cycle operating margin, asset-light. ROE is not currently meaningful (trough net income). Mid-cycle ROIC comfortably exceeds cost of capital — the signature of a real moat.

How profitable is the industry — competitors, barriers? FPGAs are a high-margin oligopoly (60–70%+ gross margins) defended by software-toolchain lock-in and long design-win/qualification cycles. Few credible competitors (AMD-Xilinx, standalone Altera, Microchip, GOWIN/Efinix); high barriers to entry on tools/IP/qualification, lower on price in commoditizing China sockets.

Can the business be easily understood? Reasonably — a focused low-power FPGA maker plus a software/IP tail. The complexity is in assessing moat durability and AI-attach economics, not the business model.

Undermined by foreign low-cost labor? Not labor; the threat is foreign low-cost competitors (Chinese FPGA vendors) eroding the price umbrella in cost-sensitive sockets — a real long-run margin risk.

Do brands matter? Modestly. “Lattice = low-power leader” carries weight with engineers, but the real stickiness is toolchain lock-in and design-win incumbency, not consumer-style brand.

Nature of competition? Design-win competition — every new socket is re-competed on power/performance/price/tools/security; incumbency protects existing sockets but must be re-earned on new designs.

Customers’ switching costs? High for existing sockets (toolchain learning, on-fabric IP, requalification cost — Greenwald customer captivity); zero for new designs. Asymmetric: strong on the legacy industrial annuity, still-being-established on the new AI book.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? The intangible toolchain/IP/design-win base and brand are not capitalized — real economic assets understated. Conversely, $315M of goodwill + intangibles inflate book equity vs. ~$395M tangible book.

Off-balance-sheet liabilities? None material flagged beyond ordinary operating leases (in capital-lease debt). The pending ~$1B AMI acquisition debt is not yet on the balance sheet (closes ~Q3’26).

How conservative is the accounting? Reasonably conservative on revenue (sell-in to distributors with reserves), but sell-in distributor accounting (~84% of revenue, ~69% via two names) sits one step from end-demand — a quality flag. SBC is fully expensed (GAAP); the company guides to non-GAAP, which excludes SBC — standard but inflates the optics of “profitability.”

How CapEx-hungry? Very light — fabless; capex ~$20M, <4% of sales. Asset-light, high-FCF-conversion model.

Capital Allocation & Management

How much FCF, and how is it used? FY25 FCF ~$155M (heavily SBC-flattered; owner-FCF ~$40M this year on the one-time grant). Used for buybacks ($100M FY25 at ~$56.72 avg, well-timed) and now M&A (AMI). No dividend.

Significant acquisitions recently? Yes — the defining one: AMI, $1.65B (~8.25x sales), announced 4 May 2026, closing ~Q3’26, the largest by >20x (prior: Mirametrix ~$68M, 2021). Funded ~$1B debt + $650M equity. Strategically coherent, financially aggressive, at a peak.

Buying back shares? Yes — $100M FY25, $15M Q1’26, new $250M authorization (Dec-2025); net share count fell despite SBC. Value-accretive timing.

Issuing large amounts of stock to insiders? SBC spiked to ~22% of revenue in FY25, but this is largely a one-time CEO inducement grant (Tamer 2024 SCT ~$69.15M), not recurring; net dilution was negative (buybacks more than offset). The AMI deal issues ~5.4M shares (~4% dilution).

Compensation policy? Fact: STI = non-GAAP operating income + GAAP revenue + MBOs (equal weight, 200% cap); LTI = >57% relative-TSR PSUs + revenue-growth PSUs (gated on Gartner non-memory-semi benchmark). No ROIC/return-on-capital metric anywhere — pay is 100% levered to growth, SBC-adjusted profit, and relative TSR. Governance hygiene otherwise solid (clawback, anti-hedge/pledge, 5x CEO ownership requirement).

Motivations of management? Interpretation: growth- and TSR-maximizing (per the comp design), with a proven value-creator CEO (Tamer built Inphi). The absence of a capital-return discipline and of insider open-market buying (~0.62% ownership, zero 5-yr code-P purchases) is a mild negative on alignment.

Valuation & Market Data

ADR, MLP, or K-1 issuer? No — ordinary US common stock (NASDAQ), 1099 reporting. No K-1.

Dividend policy? None — never paid a dividend; returns capital via buybacks.

How profitable is the business? Highly, through-cycle (see above); GAAP profitability currently masked by trough + SBC.

Net income diverging from cash from operations? Massively, and favorably in direction — FY25 OCF $175M vs. GAAP NI $3M, the gap being non-cash SBC ($115.6M) and D&A. This is the SBC distortion, not a red flag in itself, but it means GAAP NI understates cash earnings while non-GAAP overstates economic earnings (SBC is a real cost). Truth is in between.

Risks & Downside

What would cause the stock to decline? A multiple de-rating from record highs on any growth wobble; AI-attach proving cyclical; competitive re-entry (Altera/China/SoC integration); AMI integration failure or leverage strain; an industrial-recovery stall; supply/margin pressure. The dominant risk is valuation × any disappointment — a negatively skewed setup.

Risk of catastrophic loss? Low-to-moderate. Real franchise, positive equity, manageable (de-levering) debt. A 40–60% drawdown on a multiple reset + cyclical air-pocket is well within historical range (−60% in 2023–25; −85% lifetime max drawdown).

Chance of total loss? Very low — profitable, cash-generative, net-positive-equity franchise with a durable core niche.

Recent News & Events

Has the business environment changed recently? Yes, materially: (1) full cyclical inflection (Q1’26 +42% revenue / +80% EPS YoY, backlog into 2027); (2) AMI acquisition announced 4 May 2026 ($1.65B, levering up ~$1B); (3) end-market mix flip to AI/data-center; (4) channel inventory normalized (<2 months); (5) the competitive landscape reshaped by the Silver Lake/Altera carve-out (Apr-2025). News flow for LSCC over the trailing weeks was dominated by sector-level semiconductor tape moves, with no company-specific negative catalyst — a quiet idiosyncratic tape against a strong sector backdrop.

Significant acquisitions? AMI (above).

Change in accounting policies? Recast end-market reporting into Compute & Communications vs. Industrial & Embedded (Q1’26) — presentation, not policy.

Recent changes — new markets, facilities, management? New CEO (Tamer, ~Oct 2024) and the strategic pivot to “companion chip / system solutions” and M&A; entry deeper into data-center/AI server attach and (via AMI) platform firmware/security software.

APPENDIX B — Source Appendix

Report date 2026-06-21. Primary sources prioritized over secondary; management commentary treated as hypothesis and validated against filings/financials. Quantitative figures reconciled to SEC filings where applicable.

Primary — SEC Filings (EDGAR, CIK 0000855658)

1. Form 10-K, FY2025 (fiscal year ended 2026-01-03), filed 2026-02-13 — primary source for business description, product families, end-market & geographic revenue, distributor concentration, foundry dependence, employees (~1,174), R&D, risk factors, balance sheet, SBC. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000855658&type=10-K
2. Form 10-K, FY2023 (ended 2023-12-30), filed 2024-02-16 — peak-year context, segment mix at peak.
3. Form 10-K, FY2024, filed 2025-02-14; FY2022, filed 2023-02-17 — multi-year trend reconciliation.
4. DEF 14A proxy, 2026, filed 2026-03-18 — executive compensation, incentive metrics (STI/LTI), CEO inducement grant context, insider ownership, governance.
5. DEF 14A proxy, 2025, filed 2025-03-20 — Ford Tamer 2024 hiring/inducement package (Summary Comp Table ~$69.15M).
6. Form 4 corpus (insider transactions), 2021–2026 — full sweep (~694 filings since Jan-2021) for open-market purchases (code P) vs. sales/grants. Result: zero code-P open-market buys. Enumerated from the EDGAR filing index.
7. Form 8-K, Q1 2026 earnings + AMI acquisition announcement, 2026-05-04 — Q1 results, Q2 guidance, AMI deal terms ($1.65B, $1.0B cash + $650M equity, ~$1B debt, ~8.25x sales, closes Q3’26).

Primary — Earnings Call Transcript

8. LSCC Q1 2026 earnings call, 2026-05-04 (company investor-relations / public transcript) — management framing of revenue trajectory, AI/server mix (~38% server, ~25% AI, ~3 FPGAs/server), channel-inventory normalization (6mo→<2mo), Q2 guide (~$185M / ~$0.44 EPS), gross-margin outlook (~69.5%±1%), AMI strategic rationale & financing, SAM-doubling ($6B→$12B), competitive commentary (Altera divestiture, supply chain). Treated as hypothesis.

Quantitative Data Sources

9. SEC EDGAR XBRL financial data — authoritative financial facts for the US filer; CIK 0000855658. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0000855658
10. Public financial-data aggregators — multi-year income statement, balance sheet, cash flow, profitability/credit/liquidity ratios, enterprise value, and valuation multiples (FY2020–FY2025), accessed 2026-06-21. Third-party aggregated data; reconciled to filings.
11. Own-history valuation percentiles (price-to-earnings / price-to-book / price-to-sales vs. the stock’s own multi-year range), as of 2026-06-18: composite 98.7th, P/E 96.8th (GAAP-distorted — disregarded), P/B 99.42nd, P/S 99.94th percentile of the stock’s own ~10-year range.
12. Daily price history (split/dividend-adjusted) — split/dividend-adjusted OHLCV, 21/50/200 EMAs, beta (~2.31), full history; basis for the Five-Year Event Map and 52-week/5-year range. Accessed 2026-06-21.
13. Quantitative factor model — stock loadings (market beta ~1.60 all-factor / ~2.31 raw; Semiconductors +1.13; Quality +0.28; LowVolatility −0.87; no active Momentum), leaderboard (y1 return +201%, m3 annualized ~+805%, lifetime 17.6% ann / −85.3% max drawdown), stock-info (rs_12m 214, idio context), specific volatility (~41.1% annual), related stocks (MCHP, ON, NXPI, ADI, ENTG, DIOD et al.). Accessed 2026-06-21. Third-party statistical estimates; overlay only.

Public Industry / Competitive Context

14. AMD–Xilinx acquisition (closed Feb 2022, ~$49B) — public M&A record; basis for low-end deprioritization interpretation.
15. Intel–Altera carve-out / Silver Lake majority stake (announced April 2025, ~$8.75B valuation, 51%) — public M&A record; basis for the standalone-Altera competitive swing factor.
16. Competitor product context — Microchip (PolarFire/IGLOO/SmartFusion), GOWIN, Efinix, Achronix, QuickLogic — public product literature and trade press for the competitive map.

Notes on Data Treatment

• GAAP P/E (~1,000x+) disregarded for valuation — distorted by trough revenue and a one-time CEO inducement SBC charge; P/S and EV/sales used as the honest own-history valuation lenses.
• SBC treated as a real economic cost (owner-FCF = FCF − SBC), with the explicit caveat that FY25 SBC is inflated by the one-time grant and normalizes lower.
• Enterprise value computed from the current price ($153.72, 18 Jun 2026) and the FY25 net-cash position; the aggregator’s year-end EV snapshot (~$10B) reflects the stale 31-Dec price ($73.58) before the stock doubled and was adjusted accordingly.
• Management guidance (server/AI mix, SAM doubling, AMI accretion) is labeled as hypothesis throughout, not evidence.