These Billionaire Investors Keep Betting on AI Infrastructure Despite Recession Fears

David Tepper, Chase Coleman, and Bill Ackman are all increasing exposure to AI-linked stocks, with Amazon, TSMC, Vistra, and Broadcom appearing across their latest 13F filings. The most notable split is Microsoft: Tepper cut his stake 82%, Coleman reduced it 54%, while Ackman opened a new $2.1 billion position. The article is mainly a directional read on institutional positioning rather than a direct fundamental catalyst, so the likely market impact is limited.

Analysis

The important signal here is not that AI is still attracting capital, but that the market is fragmenting into distinct layers of the stack. The concentrated buying of TSMC, AMAT, AVGO, and MU suggests the more durable monetization is shifting toward the picks-and-shovels layer, where demand is less exposed to model hype cycles and more tied to capex budgets already committed for 12-24 months. That backdrop should support foundry, equipment, and memory pricing power even if end-demand for consumer-facing AI features remains noisy.

Microsoft being simultaneously cut by some managers and initiated aggressively by another tells us the market is still debating whether AI capex is creating operating leverage or simply depressing near-term FCF. The next leg likely depends on whether Azure growth can re-accelerate faster than depreciation and power costs, because if it cannot, the stock can underperform on multiple compression even while absolute revenue grows. Alphabet’s mixed treatment is a warning that investors may be skeptical of AI monetization at the application layer until ad economics or cloud margins show clear step-up.

Vistra is the overlooked second-order beneficiary: power demand is the bottleneck, not just chips. If data center load continues to outrun grid additions, utility-like assets with merchant or hybrid exposure can see a longer-than-expected earnings tailwind, but the trade is vulnerable if power prices normalize or if hyperscalers slow incremental buildouts. The risk to the whole basket is not a collapse in AI interest; it is a pause in hyperscaler capex cadence over the next 1-2 quarters, which would hit equipment and memory first and cloud platforms later.

The contrarian view is that the consensus is still chasing the wrong beta. The crowded long is not Nvidia alone but the entire AI supply chain, while the underappreciated short is any company whose valuation assumes AI spend ramps without interruption. That makes this a better relative-value environment than a simple index-long call.

Key Decisions for Investors

• Long TSM / short GOOGL for 3-6 months: own the monetization layer with more visible capex linkage and hedge the risk that application-level AI remains a narrative asset. Target 1.5x downside capture on the short if cloud/AI monetization skepticism persists.
• Buy AMAT and AVGO on 3-5% pullbacks and pair against MSFT: equipment and networking names should keep pricing power as long as hyperscaler capex stays elevated, while MSFT is the cleaner valuation-risk leg if capex scrutiny returns.
• Go long VST as a 6-12 month power bottleneck expression, but size modestly: the upside is tied to sustained data-center load growth; cut quickly if merchant power curves flatten or utility regulators signal tighter returns.
• Use MU as a tactical long into any AI-capex scare: memory typically discounts a slowdown fastest, but if buildout momentum continues, the stock can outperform sharply on inventory/pricing inflection. Favor 1-2 quarter horizon, tight stop below prior cycle support.
• Avoid chasing MSFT here; if already long, consider call spreads instead of stock for the next earnings cycle to limit multiple-risk while preserving upside if Azure growth outpaces capex concerns.