Is Billionaire Michael Platt Souring on the Magnificent Seven? He Just Closed Positions in 4 AI Titans in Favor of a Company That May Be a Surefire Winner in the AI Infrastructure Spending Boom.

Michael Platt’s BlueCrest sold all of its Nvidia, Microsoft, Meta, and Amazon positions in Q1, while increasing Taiwan Semiconductor Manufacturing (TSM) by more than 430% to 35,824 shares, now 0.7% of portfolio value. The article argues TSMC may be a key beneficiary of AI infrastructure spending, given its role manufacturing chips for Nvidia, Broadcom, and AMD. The piece is largely analytical commentary on hedge fund positioning rather than a direct company catalyst.

Analysis

This looks less like a verdict on AI and more like a rotation from hyperscaler narrative risk into the higher-beta toll booth on the value chain. If capital spending stays elevated, the economically superior exposure is not the end-demand platform but the foundry that monetizes every custom ASIC, GPU, and networking ASIC regardless of winner. That makes TSM the cleanest way to express continued AI infrastructure intensity while reducing single-name model risk versus the Mag 7.

The second-order effect is important: the more AI capex becomes a procurement war, the more pricing power shifts to the constrained manufacturing layer. That should support not only TSM but also adjacent enablers like AVGO and, to a lesser extent, AMD if incremental accelerator share keeps broadening. By contrast, the large platform names face the risk that investors increasingly demand proof of incremental ROI from AI spend, which can compress multiples even if fundamentals remain fine.

The near-term catalyst is any evidence of capex re-acceleration or supply tightness in advanced nodes, packaging, and HBM-related demand. The main reversal risk is a digestion quarter: if cloud buildouts pause for 1-2 quarters, TSM can still grow, but the market may de-rate the whole infrastructure complex before fundamentals catch up. That creates a window where the trade works operationally but not necessarily immediately in price.

Contrarian read: the crowd is still treating AI as a winner-take-all equity story, when the better risk-adjusted trade may be picking the bottleneck, not the brand. The article’s signal is likely underappreciated because it implies that even sophisticated allocators are moving from equity narrative exposure into picks-and-shovels exposure. If that becomes a broader pattern, TSM/AVGO should outperform on relative earnings revisions even if the Mag 7 continues to grind higher.

Key Decisions for Investors

• Go long TSM vs. a basket of NVDA/MSFT/META/AMZN on a 3-6 month horizon; thesis is that market leadership shifts from AI beneficiaries to AI infrastructure monetizers, with better earnings visibility and less multiple risk.
• Add to TSM on pullbacks, preferably after any cloud-capex digestion headline; target a 12-month horizon with asymmetric upside if advanced-node utilization stays tight, but cut if capex guidance rolls over for two consecutive quarters.
• Initiate a modest long AVGO/short NVDA pair for 1-2 quarters; if AI spend broadens beyond GPUs into custom silicon and networking, AVGO should capture more durable mix, while NVDA remains more sentiment-sensitive.
• For event-driven exposure, buy TSM calls or call spreads into the next earnings cycle; risk/reward improves if implied volatility underprices confirmation of AI-related demand persistence.
• Avoid chasing the recent Mag 7 tape at current levels; use any further multiple expansion in GOOGL/AAPL/TSLA as a funding source for higher-conviction infrastructure names.