Third Point LLC just filed a 13F revealing a significant reduction in its Lam Research stake. The market shrugged. The crypto crowd barely noticed. That's a mistake.
This isn't a semiconductor story. It's a narrative shift about the capital expenditure cycle that powers the entire AI infrastructure stack—from HBM memory to GPU fabrication to the energy grids that run your proof-of-work and proof-of-stake networks. When a hedge fund known for macro precision exits a key equipment supplier, the signal ripples through every layer of the crypto economy.
Let me decode the signal from the blockchain noise.
Context: The Lam Research Position in the Crypto AI Stack
Lam Research is not a household name in crypto. It should be. The company dominates the production of etch and deposition equipment used to manufacture HBM (High Bandwidth Memory) and advanced logic chips. Every NVIDIA H100 or B200 GPU that secures the AI training layer—and by extension the AI tokens, DePIN projects, and zk-proof accelerators built on top—relies on Lam's tools for the TSV (through-silicon via) etching that makes HBM stacks possible.
Third Point's Dan Loeb is not a crypto native. He's a value-oriented activist. But his firm's move to sell Lam Research at a time when the AI narrative is at peak euphoria should make every crypto investor pause. The narrative hunters in crypto are currently chasing the ghost of 2017's fever dream, piling into anything with 'AI' in the ticker. Loeb is selling the shovel provider.
In my 24 years of observing market cycles, the moment the smart money starts rotating out of the picks-and-shovels plays is the moment the retail herd is about to get trampled.
Core: The Seven Dimensions of the Lam Signal
1. Technology Cycle: The HBM Equipment Cliff
Lam's technology moat in high-aspect-ratio etch for 3D NAND and TSV for HBM is undeniable. The company holds 40%+ market share in critical etch and deposition steps. But the equipment cycle is not linear. The peak of HBM capacity expansion—driven by the hyperscaler AI capex boom—is likely behind us. The next phase is incremental, not explosive.
Structuring chaos into profitable narratives requires understanding that the highest growth rate is always in the early innings. Once HBM production lines are built, the equipment orders slow. Lam's FY2025 guidance will be the first real test. If Third Point is selling now, they are betting that the Q2 2024 earnings beat was the peak.
2. Supply Chain: The Export Control Trap
Lam's revenue from China dropped from 29% in FY2021 to ~20% in FY2023. The US export controls are not temporary. They are a structural headwind. The Chinese semiconductor ecosystem is accelerating domestic substitution—companies like AMEC and NAURA are eating into Lam's mature-node market share. Meanwhile, the 'friendshoring' builds in the US, Japan, and Europe are slow and expensive.
For crypto, this means that the supply chain for AI chips is becoming more fragmented and less efficient. The cost of GPU manufacturing is rising, which will eventually trickle down to the cost of securing AI-driven networks. The narrative of 'cheap AI compute' is a mirage.
3. Capacity Cycle: The Leading Indicator of Crypto Mining Hardware
Surviving the winter to harvest the spring—I've seen this movie before. In 2021, when Ethereum mining GPU prices peaked, the leading indicator was the order book of semiconductor equipment companies. Lam's orders lead wafer fab capex by 12-18 months. Third Point's sale is a bet that the global wafer fab equipment market will contract from $100B+ in 2025 to below $90B in 2026. That means fewer new fabs, less GPU capacity, and potentially higher prices for ASICs and GPUs used in crypto mining.
4. Demand: The AI Capex Marginal Growth Slowdown
Hyperscalers spent over $200B on AI capex in 2024, with 30%+ growth expected in 2025. But the marginal growth rate is decelerating. The equipment companies are the first to feel the pain because they are the most leveraged to the 'acceleration' phase. Lam's stock trades at 30-35x PE, well above its historical average of 25x. The AI narrative premium is priced in. The moment the market realizes that AI chip revenue growth is outpacing AI equipment revenue growth by a factor of 2x, the equipment stocks will re-rate.
For crypto investors holding AI tokens like Render, Akash, or Bittensor, this is a canary in the coal mine. If the physical infrastructure for AI compute stops expanding at the same rate, the tokenomics of these networks—which rely on continuous hardware onboarding—will face headwinds.
5. Geopolitics: The Decoupling Tax
Alpha isn't extracted from narratives; it's extracted from structural inefficiencies. The US-China tech decoupling is creating a massive inefficiency in the semiconductor supply chain. Lam is caught in the crossfire—unable to serve the Chinese market for advanced tools, while facing increased competition from Chinese domestic players. The 'decoupling tax' will eat into their margins and growth rates for the next 3-5 years.
This has a direct implication for crypto: the narrative of 'decentralized AI' as a geopolitical hedge becomes more relevant. If the US restricts access to the best AI hardware, non-US developers will turn to permissionless compute networks. Projects like Akash and io.net could see a tailwind, but only if they can secure the hardware—which is becoming harder to source.
6. Competition: The Market Share Battle
Lam's competitive moat is real but narrowing. Applied Materials is investing heavily in hybrid bonding and advanced packaging, directly threatening Lam's HBM equipment monopoly. Tokyo Electron is winning share in Korean memory fabs. The market share stability Lam enjoyed in the 2010s is eroding. In a scenario where the total addressable market shrinks, the competitive intensity increases. The 'picks and shovels' are no longer a sure bet.
7. Valuation: The 'Sell in Hype, Buy in Fear' Play
Third Point's exit is a textbook example of selling when the narrative is strongest. Lam's PE ratio of 30-35x is unsustainable if the growth rate drops from 15% to 8%. The fund is rotating into assets with more asymmetric upside—like pure AI compute plays (NVIDIA) or event-driven situations. For crypto, the lesson is clear: when the hedge funds that drove the AI narrative start selling the equipment providers, the narrative itself is approaching exhaustion.
Decoding the signal from the blockchain noise—the real story is not about Lam's fundamentals. It's about the end of the first phase of the AI infrastructure buildout. The second phase will be different: less capital-intensive, more software-driven, and more focused on efficiency. The tokens that survive will be those that align with that shift.
Contrarian: The Crypto Market Is Misreading the Signal
The conventional crypto take will be: 'Third Point selling Lam means AI is overvalued, so sell all AI tokens.' That's lazy. The contrarian truth is more nuanced.
Third Point is not selling because AI demand is collapsing. They are selling because the equipment cycle is peaking. The equipment cycle is a leading indicator for hardware supply, not for AI software demand. The hyperscalers will continue to build out data centers, but the rate of capacity addition will slow from 'exponential' to 'linear'. That's a rotation from growth to value, not a crash.
For crypto, this means:
- The GPU shortage for AI training will ease, making it cheaper for new projects to acquire compute. That's bullish for decentralized AI networks that rely on spare capacity.
- The HBM supply glut that will follow the equipment cycle peak will lower the cost of memory for high-performance nodes. That could make zk-proof hardware cheaper.
- The narrative shift from 'AI infrastructure' to 'AI application' will reward tokens that focus on utility, not speculation.
The illusion of value in digital scarcity—the crypto market attaches value to scarcity. But Lam's equipment is not scarce. The cycle will produce excess capacity. The scarcity is in the application layer, not the hardware layer.
Takeaway: The Next Narrative Shift
Watch the hyperscaler capex guidance for Q1 2025. If Microsoft, Amazon, and Google all guide for a deceleration in AI infrastructure spend, the equipment stocks will follow Third Point out the door. That will be the signal to rotate from AI infrastructure tokens (compute, storage) to AI application tokens (agents, data, inference).
If they maintain or increase guidance, Lam's dip is a buying opportunity for the next 12-month cycle. But the probability is low. The marginal dollar is already moving from hardware to software.
History doesn't repeat, but it rhymes. The 2017 ICO boom saw the shovels (Ethereum, miners) peak before the applications (decentralized finance). The same pattern is playing out in AI. The equipment sells first. The tokens follow. The question is: are you still holding the shovel when the ground starts shaking?