David Tepper’s Appaloosa nearly doubles Amazon stake, adds Sandisk in the first quarter

David Tepper’s Appaloosa Management sharply increased its AI/tech exposure in Q1, making Amazon its largest disclosed holding at roughly $900 million after a 98% position increase. The fund also boosted Uber by 242%, Vistra Energy by 114%, and added to Taiwan Semiconductor and Micron, while initiating a new Sandisk stake valued at about $179 million. Offsetting that, Appaloosa cut Alibaba by 33%, Alphabet by 3%, and Nvidia by 13%, but all three remained among its 10 largest U.S. equity holdings.

Analysis

This is less a generic AI chase and more a clear preference for the picks-and-shovels that monetize inference, power, and distribution with lower single-name execution risk. The biggest second-order winner is not the headline semiconductor complex but the ecosystem around data-center load growth: hyperscaler capex, grid equipment, gas-to-power and merchant power pricing all get a bid when a large fundamental investor reinforces the “AI = electricity + memory + cloud” trade. That matters because it broadens the rally beyond the narrow GPU bottleneck and makes the trade more durable if hardware supply constraints ease.

The asymmetric read-through is that Tepper is expressing more confidence in monetization durability than in peak chip beta. The trim in the highest-multiple AI beneficiary versus adds in cloud, mobility, memory, and energy suggests a shift from pure narrative exposure to cash-flow exposures with more visible near-term catalysts. That is bullish for names with operating leverage to AI adoption but also with a real-world demand driver over the next 6-12 months, especially where consensus underestimates incremental power and storage intensity.

The main risk is that this trade works only if AI capex remains broad-based rather than rotating into a smaller set of preferred platforms. If enterprise spend stalls, the memory and networking chain will usually de-rate first, then cloud-adjacent winners follow with a lag of one to two quarters. A sharper risk-off in mega-cap tech would likely pressure the whole basket, but the relative losers should be the names with the most crowded ownership and least pricing power.

Contrarian angle: the market may be overconfident that every AI spend dollar is equally durable. Memory and storage can look like “AI beneficiaries,” but historically they are the first segment where buyers negotiate aggressively once supply tightens and pricing inflects; that makes the payoff path more cyclical than structural. Tepper’s moves argue for owning the revenue attached to AI adoption, not just the compute story itself.

Key Decisions for Investors

• Go long AMZN vs short NVDA in a 3-6 month pair: favor monetization and demand diversification over the most crowded AI multiple; target 10-15% relative outperformance if AI sentiment stays positive.
• Initiate a long SNDK / MU basket on a 1-2 quarter horizon, but size modestly and hedge with a tech-beta short; upside is cyclical margin expansion, downside is rapid pricing compression if channel inventories rebuild.
• Add UBER on pullbacks over the next 4-8 weeks: this is a lower-consensus AI-adjacent beneficiary through mapping, automation, and ad-tech optionality, with cleaner operating leverage than hardware names.
• Use call spreads on AMZN into the next earnings cycle: the position benefits from continued cloud and AI demand while capping premium outlay if the market rotates away from mega-cap growth.
• Avoid chasing NVDA here; if long at all, pair it against a basket of under-owned AI beneficiaries to reduce multiple risk, since the marginal buyer appears to be getting more selective.