Discovery Capital Management Llc Ct Rotates Hard Into AI Semis and LatAm Risk

The biggest adds read like a curated bet on the picks-and-shovels of the AI cycle and the under-banked consumer in Latin America. New and upsized Tech positions dominate the capital deployment, but they’re flanked by bold moves in EM financials and high-operating-leverage cyclicals.

• SNDK at 6.06% is the single largest line in the book, a new $116.7M bet on high-value memory/flash capacity, with the position already up 222.6% vs their $436.36 average buy. That signals comfort owning volatile, structurally advantaged capacity in the AI buildout rather than just the headline GPU names.
• ON comes in fresh at 4.53% and $87.3M, a big statement that power and analog semis will remain tight as AI and EV demand collide. The nearly 95.0% gain vs a $58.03 cost suggests they are willing to add even after a major run.
• MU is pushed to 4.85% after a +162.4% share add worth about $57.7M, a classic “double-down as the thesis works” move on AI memory cycles.
• LRCX sees a near‑doubling of shares (+96.2%, +$31.2M), underscoring conviction that semicap tools are the long-duration winners from capex plans, not just one-node chip makers.
• IFS is a new $53.1M stake at 2.76% alongside large increases in NU (+424.6% shares, +$26.1M) and BBAR (+124.8% shares). Coupled with bigger positions in BAP and GGAL, this is a clear thematic bet on Latin American retail and commercial credit penetration.
• CAR enters at 4.16% and $80.0M, while AMZN is ramped aggressively (+187.3% shares, +$16.9M). Along with new ULTA, new ROST, and a larger GENI stake (+79.0% shares), Discovery is leaning into a consumer upcycle where discretionary spend, travel, and digital ad dollars re-accelerate.
• IBIT’s position explodes (+9331.9% shares, +$24.5M) while IREN is trimmed, signaling a pivot from single-name crypto miners to more liquid, diversified Bitcoin exposure.

Layer on new NVDA, FSLR, TER, CRM, INTU, and SOXX, and you get a portfolio that’s intentionally closer to the AI and digital infrastructure beta than it has been in prior quarters.

The sales side confirms that Discovery isn’t de‑risking; it’s re‑risking away from old hedges and toward new themes. The biggest dollar trims hit prior home-run trades, industrial cyclical exposure, and defensive staples.

• IREN is cut by 29.0%, freeing up about $25.7M after a 479.9% gain vs their entry. That’s textbook risk recycling: take capital out of a hyper-volatile crypto miner and redeploy into IBIT, AI semis, and LatAm banks.
• JPM is reduced by 31.9%, roughly $21.3M, even though the line is up 19.2% vs cost. In the context of big adds to NU, IFS, BAP, GGAL, and BBAR, this looks like a switch from mature U.S. banking spread trades to higher-growth, cheaper EM financials.
• PRMB and JBS both give ground, with PRMB cut 19.4% (about -$6.3M) and JBS still being averaged into despite being slightly underwater. Net, Consumer Staples exposure falls as the fund exits the “food and drink as inflation hedge” trade.
• CLF sees a 26.8% reduction (about -$4.5M), and AEM is trimmed 14.1%. Combined with leaving PPTA flat, this signals a willingness to dial back metals and gold just as the book leans harder into growth and EM credit.
• Smaller trims in VNET and CHDN look like housekeeping: crystallizing some gains in volatile China data infrastructure and a challenged U.S. gaming/leisure name to fund higher-conviction growth ideas.

Importantly, there are no evident wholesale evacuations from structurally broken theses. This quarter’s sells are about shifting the risk budget, not retreating from markets.

The sector bars tell a clean story: Discovery is methodically exchanging mature cyclicals and defensives for growth, infrastructure, and risk assets. Technology’s jump to 33.6% and Consumer Discretionary’s move to 16.71% come straight out of Energy, Industrials, Telecom, Consumer Staples, and Basic Materials.

Energy drops from 9.09% to 7.14% despite new money into VLO and LNG, because the rest of the portfolio is growing faster around it. Industrials fall to 6.56% as they lean away from traditional capex proxies like building materials and legacy broadcasters, and toward semicap (LRCX, TER) inside the Tech bucket.

Telecom exposure via AMX shrinks from 10.17% to 6.17% in portfolio terms despite a modest add, reflecting a view that EM telecom is no longer the core EM growth proxy; banks and fintechs are. Consumer Staples slide from 9.64% to 5.77%, and Basic Materials from 5.92% to 3.15%, as metals, gold, and food hedges are sold to fund AI and fintech.

At the margin, they also introduce new “risk sleeves”: an explicit Bitcoin allocation via IBIT and a dedicated semiconductor ETF via SOXX, now grouped under Unclassified at 2.14%. Utilities and Health Care appear via LNG and MDLN, but those are small, idiosyncratic bets, not a defensive tilt.

Pulling it together, Discovery’s 2026‑Q1 13F looks like a manager consciously trading away ballast for upside volatility. After a -1.5% latest-quarter weighted performance against a strong 3‑year track record, they are not hiding in cash-flow stability; they are chasing where they think multi-year earnings power is mispriced.

The AI stack is now central: memory (SNDK, MU), power and analog (ON), tools (LRCX, TER), GPUs (NVDA), and renewables capacity (FSLR), plus software layers (APP, CRM, INTU, PINS). On top of that sits a discrete Bitcoin sleeve (IBIT), suggesting they see digital assets as a structural asset class rather than a trade through miners alone.

In parallel, the book is reoriented around EM household balance sheets: IFS, NU, BAP, GGAL, BBAR, and RKT speak to a belief that Latin American and U.S. subprime/alt-mortgage credit growth can compound far faster than U.S. megabanks. Consumer cyclicals (CAR, AMZN, ULTA, ROST, GENI, GLNG) round out a view that real disposable income and travel/leisure demand still have room to run.

If this is right, future quarters should show continued heaviness in Tech and EM financials, with occasional trims in the big AI winners to fund new risk pockets. If they’re wrong, the portfolio will feel the drawdown: this is a book set up to outperform in an AI- and EM-led expansion, not one designed to sidestep a global growth scare.