Stanley Druckenmiller Backs These 3 AI Infrastructure Stocks: Should You Follow?

Duquesne Family Office disclosed themed AI-infrastructure positions in Broadcom, Seagate, and Micron in its Q1 2026 13F, validating the trade at the portfolio level. Broadcom reported $22.19B revenue, up 47.9%, with AI semiconductor revenue of $10.80B (+143%) and guided Q3 AI revenue to grow over 200% to $16.0B. Seagate posted 44.1% revenue growth to $3.11B with gross margin expansion to 47.0%, while Micron delivered $23.86B revenue, up 196.3%, and guided Q3 revenue to $33.50B with roughly 81% gross margin.

Analysis

The common thread is not “AI winners” so much as the re-rating of the entire upstream bottleneck stack. If hyperscalers keep converting capex into software-driven demand, the first second-order effect is that component suppliers with operating leverage become quasi-levered bets on AI capex durability; that benefits AVGO most because its custom-silicon franchise is tethered to multi-year design cycles, while MU and STX are more exposed to buyers pausing orders if growth expectations wobble. The market is implicitly telling us that AI is moving from experimental spend to infrastructure spend, which usually compresses the time between orders and margin inflection — but it also increases the penalty if capex growth normalizes even modestly.

The biggest hidden risk is that the three names sit at very different points on the cycle despite being grouped under one theme. MU’s earnings power can look almost too strong for the current tape, but memory is the easiest place for customers to over-order into scarcity and then correct hard 2-4 quarters later; that makes the forward multiple look cheap until it suddenly isn’t. STX has the best near-term margin narrative, but storage tends to lag compute spending by a few quarters, so the market may be pricing a continuation of AI build-out that is already partly reflected in cloud budgets.

AVGO is the most defensible long-duration compounder because software attaches to the silicon and lowers earnings volatility, but even here the stock’s valuation assumes the AI mix keeps compounding without a digestion phase. The contrarian miss is that consensus may be underestimating how much of the upside is already being pulled forward by crowded ownership and thematic chasing. In other words, the thesis is likely right over 1-3 years, but the path matters: the next earnings or capex comment could easily produce a “good news, bad reaction” setup if guidance merely meets elevated expectations.

Key Decisions for Investors

• Overweight AVGO on a 6-12 month horizon; best risk/reward in the group because recurring software revenue can cushion any AI semiconductor digestion, but size it below a core mega-cap weight given crowded thematic positioning.
• Use MU as a tactical trade, not a strategic hold: buy weakness only after a post-earnings vol reset or on a 10-15% pullback, because memory upside is most vulnerable to a 1-2 quarter demand pause.
• Fade STX into strength via a partial short or put spread if it re-rates further; the stock is the most exposed to mean reversion if cloud customers slow storage purchases after current AI buildouts.
• Pair trade: long AVGO / short STX for the next 2-4 quarters, betting that software-backed custom silicon has more durable multiple support than storage, which is more cyclical and more crowded.
• If running options, prefer call spreads over outright longs in MU or STX; implied volatility likely understates the risk of a sharp post-guidance compression after a crowded AI tape.