LAM RESEARCH CORP

LAM RESEARCH CORP (LRCX) — Investment Memo

🐂 The Bull Case (Warren's voice)

• The moat is physics, not marketing. Once a fab qualifies a Lam etch/deposition tool for a node, swapping requires months of requalification and yield risk. Switching costs are absolute. This isn’t a subscription service – it’s embedded in the process recipe. The service annuity grows with every new tool shipped, compounding revenue without incremental sales effort.
• Economics are exceptional – and getting better. Gross margins expanded from 15.5% (2016) to 29.1% (2025). ROE went from 15.5% to 54.3% – not leverage magic (debt stable at ~$4.5B), but genuine pricing power and mix shift toward high-margin service. Net income $5.4B covers interest 12x over. Even in a cyclical trough (2024 revenue $14.9B), they earned $3.8B net – a 25% margin. The machine prints cash when the cycle is merely bad.
• Attractive entry price? Only if the market prices in a permanent recession. Normalized earnings power (mid-cycle revenue ~$16B, net margin 25%) = $4B net. At 15x (fair for a oligopoly with switching costs), intrinsic value is $60B – roughly $460/share (assuming ~130M shares). If the stock dips to $345 (25% discount) or $230 (50% distress), Berkshire should buy aggressively. At today’s price? We need to check. The moat survives 20 years. The price must be the margin of safety.

🐻 The Bear Case (Charlie inverts)

• Scenario #1: Technological bypass. Atomic layer deposition (ALD) or a completely new etch method (e.g., self-assembled patterns) that renders plasma-based tools obsolete. If a competitor like Applied Materials or Tokyo Electron delivers a 10× cheaper, faster solution that requires requalification, Lam’s installed base becomes a stranded asset. Likelihood? Low over 10 years – chip fabs move slowly. Over 20 years? Real. Lam’s R&D response will be frantic.
• Scenario #2: Customer concentration kills the cycle. The 10-K discloses two customers each >10% of revenue (likely Samsung and TSMC). If one of them vertically integrates its own tool development (Samsung has the capability) or decides to shift process steps to a Lam rival, revenue collapses by 20-30% overnight. Switching costs protect Lam, but a determined customer with deep pockets can overcome them given 3-5 years of engineering effort.
• Scenario #3: The hidden cash cancer. No free cash flow disclosed in the 10-year record. SBC is aggressive – share count down ~30% but at a cost: stock-based compensation inflates net income. If true FCF trails net income by 15-20% (typical for semi-cap equipment due to working capital swings), then $5.4B net income is really $4.3-4.5B cash. On a normalized basis, the “earnings” multiple is actually 18-20x, not the 12x it appears. This is the most likely permanent impairment – not a crash, but a slow bleed of overvalued equity.

💰 Valuation & Margin of Safety

• Intrinsic value estimate: $460/share (normalized net income $4B, 15x multiple, ~130M shares). This assumes mid-cycle revenue $16B, 25% net margin, and no structural disruption. The bull case says $600 (peak earnings x 15); the bear says $300 (normalized FCF at 12x).
• 25% margin of safety entry: $345/share – where a buyer gets a 33% normalized earnings yield while accepting cyclical risk.
• 50% margin of safety entry: $230/share – Buffett’s ideal. At that price, the business is priced for permanent disaster. Even if one of the three bear scenarios materializes, you still earn a decent return.
• Currently: No price given, but based on the peak-cycle multiples and missing FCF, LRCX likely trades above $460. That means it is expensive – a cyclical high masquerading as a consistent compounder. Wait for the next semiconductor downturn (it will come).

Verdict: WATCH

The business has a genuine, widening moat and superb economics – but the lack of cash flow transparency, peak-cycle margins, and hidden SBC dilution mean the current price likely offers no margin of safety. Buy only at a 25-50% discount to normalized intrinsic value; until then, let the cycle bury the hype.