The filing hit the SEC wire at 4:17 PM on a slow Tuesday. Third Point LLC, the activist fund run by Daniel Loeb, had dumped its entire stake in Lam Research. No fanfare. No press release. Just a 13G filing buried in the EDGAR database. The market barely blinked. Lam shares traded sideways for three days, then started drifting lower. The move was a tactical portfolio adjustment, analysts said. Nothing to see here. They were wrong.
This is what happens when a fund with a 30-year track record of timing semiconductor cycles exits a position without explanation. The market interprets absence as noise. But for those who read the filing as a forensic document, the signal is unmistakable. Third Point's exit from Lam Research is not a bet against the company. It is a bet against the current phase of the capital expenditure cycle. And that is a much more dangerous trade to face.
Context: The Sell Signal in the Machine
Lam Research is not a speculative crypto project with a whitepaper and a promise, but the analytical framework is the same. Audit the code, not the pitch. In this case, the code is the WFE (wafer fab equipment) order book, the capital expenditure commitments of the world's largest foundries, and the regulatory sand traps that the US government has laid around the entire industry.
Lam is the dominant supplier of etch and deposition equipment for advanced memory and logic manufacturing. Its tools are essential for 3D NAND stacking, HBM TSV (through-silicon via) etching, and the GAA (gate-all-around) transistor architectures that will power the next generation of AI chips. The company holds roughly 20% of the global WFE market, with a particularly strong position in the memory segment, where its etch-deposition combo tools have a market share exceeding 40%.
Financial profile: FY2023 revenue of $17.4 billion, down 25% year-over-year due to the memory downturn. FY2024 expected to recover to $15-17 billion. Gross margins in the 44-46% range. Net margins around 22-24%. Cash rich, low leverage, and a consistent return of capital through buybacks and dividends. On paper, this is a fortress. So why would one of the market's most respected value-oriented funds walk away?
Core: The Systematic Teardown
Let me reconstruct the logic that led to this decision. It is not a single factor. It is a convergence of four structural shifts that together create a risk profile that no longer fits the portfolio.
First: The China Revenue Cliff. Before the 2022 export controls, Lam derived roughly 29% of its revenue from China. By FY2023, that number had dropped to 20-25%. The trajectory is clear: the US Bureau of Industry and Security (BIS) has adopted a presumption of denial for advanced process equipment to China. This covers the exact product lines where Lam's technology is most differentiated: sub-16nm logic etch, 128-layer+ 3D NAND equipment, and 18nm half-pitch DRAM tools. The company can still sell service and spare parts into China, which carry higher margins (60%+), but the volume of new equipment sales is structurally constrained.
Here is the math that most analysts miss. The China revenue decline is not just about lost sales. It is about the loss of the incremental growth engine. The CHIPS Act in the US, the European Chips Act, and the Japan Semiconductor Revival Plan are all pouring money into fab construction. But these are offset markets. They are replacing, not expanding, the global demand base. The only genuine growth market for semiconductor equipment over the next five years is China's domestic buildout. And Lam is being systematically excluded from it.
Second: The AI Capital Expenditure Deceleration Risk. The bull case for Lam, and for all WFE stocks, rests on a single assumption: that AI-driven capital expenditure by cloud service providers (CSPs) will continue to grow at 30%+ annually for the next three years. This assumption is embedded in the stock's valuation. At 30-35x trailing earnings, Lam is priced for a level of demand that assumes linear extrapolation of the current AI infrastructure buildout.
But the reality of capital expenditure cycles is that they are lumpy, not linear. The major CSPs—Amazon, Microsoft, Google, and Meta—collectively spent over $200 billion on AI infrastructure in 2024, with a projected 30%+ increase in 2025. But equipment orders are a leading indicator of fab construction, not a lagging one. The order books for Lam, Applied Materials, and Tokyo Electron are already reflecting the peak of the current cycle. Third Point's exit is a bet that the next 12-18 months will see a deceleration in the rate of growth of WFE spending, not a collapse, but a moderation that will compress the valuation multiples of the entire sector.
Third: The HBM Equipment Peaking Dynamic. High Bandwidth Memory (HBM) is the hottest segment in memory, driven by the need for high-speed memory in AI accelerators. Lam is a critical supplier of the TSV etch and deposition equipment used in HBM manufacturing. The HBM equipment market grew roughly 50% in 2024 and is expected to grow another 30-40% in 2025. But here is the structural flaw: as HBM production matures and the technology transitions from HBM3 to HBM4, the equipment intensity per wafer declines. The process becomes more standardized, the tool-to-tool variation diminishes, and the average selling price (ASP) of the equipment softens. This is a classic pattern in the semiconductor industry. The first wave of a new technology creates a step function in equipment demand. The second wave creates a plateau. Lam is currently riding the first wave. The smart money is already positioning for the plateau.
Fourth: The Regulatory Sand Trap. The US export control regime is not static. It is evolving in a direction that is increasingly hostile to the business model of US-based equipment companies. The 2022 and 2023 rules were the first salvo. The 2024 rules, which include restrictions on the service of advanced equipment in China without a license, represent a qualitative escalation. The risk is not just that Lam loses China revenue. The risk is that the compliance burden, the legal uncertainty, and the geopolitical friction permanently impair the company's ability to serve its global customer base. The cost of doing business is rising, and the return on invested capital is declining.
Contrarian: What the Bulls Got Right
I will not pretend that the bear case is the only valid case. The bulls have several strong arguments. First, Lam's technology moat is real. The combination of high-aspect-ratio etch and atomic layer deposition (ALD) is not easily replicated. The Chinese domestic champions, AMEC (Advanced Micro-Fabrication Equipment) and Naura, are making progress in mature nodes, but they remain years behind in the advanced process equipment that Lam dominates. The customer switching cost is enormous. Once a fab is qualified with Lam's tools, the cost of requalifying with a competitor's equipment is measured in quarters, not weeks.
Second, the AI demand story is not a bubble. The $200 billion+ of CSP capital expenditure is real, and a significant portion of it is flowing into the advanced packaging and HBM capacity that Lam serves. The long-term trend is structural. The question is only about timing and magnitude.
Third, Lam's financial strength is undeniable. The company generates over $3 billion in free cash flow annually, maintains a fortress balance sheet, and returns capital to shareholders through buybacks and dividends. Even if the stock trades down 20% from current levels, the dividend yield and buyback yield provide a meaningful floor.
Takeaway: The Accountability Call
Third Point's exit from Lam Research is not a warning about the company's technology or its competitive position. It is a warning about the cycle. The semiconductor equipment industry is a classic cyclical growth story. The growth is real, but the cycle is inevitable. The question every investor should ask is not whether Lam is a good company, but whether the current price embeds an assumption of perpetual growth. The answer, based on the exit signal from one of the market's most experienced cycle traders, is clear.
Audit the code, not the pitch. The code here is the order book, the export control regime, and the capital expenditure commitment timelines. The narrative is the AI revolution. Do not confuse the two. Complexity hides risk. And in this case, the risk is that the next 12 months will reveal that the equipment cycle peaked in the second half of 2024. The smart money is already positioned for the descent. The rest of the market is still admiring the view from the top.
