Billionaire Philippe Laffont Trimmed 8 of His Fund's Top 10 Positions Last Quarter. Here's What the Other 2 Have in Common.

Coatue Management shifted toward AI infrastructure names last quarter, adding to Taiwan Semiconductor Manufacturing and Lam Research while cutting hyperscalers and chipmakers like Nvidia and Broadcom. The article argues TSMC remains attractively valued at about 25x earnings versus Lam Research at 48x forward P/E, despite strong share gains. Overall the piece is a portfolio and valuation commentary suggesting continued relative strength for upstream AI supply-chain companies.

Analysis

The signal is not “own AI,” but own the bottleneck with the cleanest pricing power. When hyperscalers keep ratcheting capex, the economic rent migrates upstream to foundry and equipment vendors first, then to memory as the next-leg bottleneck tightens; that is why TSM, LRCX, ASML, and MU matter more than the headline beneficiaries. The second-order effect is that AI spend can remain structurally inflationary for the supply chain even if cloud margins compress, which argues for a longer-duration premium on the tools and wafer nodes with real capacity scarcity.

The market is starting to differentiate between durable scarce assets and “good businesses at bad prices.” NVDA and AVGO still have exceptional franchises, but if customers increasingly negotiate against them via TSMC and the equipment ecosystem, their bargaining power and multiple expansion may be capped relative to the picks-and-shovels names. That makes the current setup less about absolute AI exposure and more about relative positioning within the stack, with upstream names likely to outperform on both earnings revisions and multiple stability over the next 6-12 months.

The contrarian issue is that the consensus may be underestimating how much of this good news is already embedded in the valuation of the strongest upstream names. LRCX and ASML have become crowded “must-own” AI proxies, so any pause in capex growth, export-control friction, or a digestion quarter in memory could trigger fast de-rating despite still-solid fundamentals. The cleaner asymmetry appears in TSM, where valuation remains more reasonable versus its strategic leverage; the risk there is geopolitical, but the next 1-2 quarters should still be driven by pricing and utilization rather than end-demand headlines.

Key Decisions for Investors

• Long TSM vs short a basket of hyperscalers (MSFT/AMZN/GOOGL/META) for 3-6 months: thesis is margin transfer upstream; target relative outperformance if capex keeps compounding faster than cloud monetization.
• Add to LRCX on pullbacks, but size as a momentum-within-valuation trade rather than a core compounder; use a 10-15% trailing stop because the stock is vulnerable to any capex hesitation after a strong run.
• Initiate long ASML / short NVDA pair over 6-12 months: both benefit from AI spend, but ASML has cleaner scarcity economics and less customer concentration risk from hyperscaler bargaining power.
• Buy TSM call spreads 6-9 months out instead of stock if available: it captures continued upstream rerating while limiting geopolitical downside from Taiwan-specific headlines.
• Reduce or hedge crowded winner exposure in AVGO and NVDA into strength; if AI spend normalizes, those names are more exposed to multiple compression than TSM/LRCX/ASML.