Citadel Advisors 13F Portfolio · Ken Griffin

The biggest adds by dollars show Citadel leaning into three themes: AI infrastructure beyond the usual suspects, platform consumer/tech, and a new pharmaceutical profit pool. The enormous reload into Eli Lilly (LLY), up +291.1% with about $844.1M added, and AbbVie (ABBV), up +546.5% with about $674.3M added, anchors a high-conviction call on obesity, immunology, and large-cap pharma pricing power.

On the AI side, they are clearly reallocating away from over-earned winners into the picks-and-shovels. KLA (KLAC) explodes +6607.7% to roughly $766.7M despite being deeply underwater at -76.8% vs cost — a rare move that signals process-based conviction, not P&L chasing. Seagate (STX) climbs +582.2% with about $622.2M added, and Texas Instruments (TXN) is up +459.3%, pointing to a long runway for memory, storage, and analog in an AI-heavy compute world.

Consumer and software platforms remain core growth funding ideas, not afterthoughts. Amazon (AMZN) gets roughly $662.5M more capital, Apple (AAPL) and Microsoft (MSFT) are up around 40–43% in shares, and Electronic Arts (EA) more than doubles. In software, Snowflake (SNOW) is up +116.2%, and Arista Networks (ANET) rises +411.9%, signaling a belief that data infrastructure and high-speed networking will monetize the AI boom more steadily than the most speculative model plays.

The funding list is dominated by one message: take money off the table where AI and cyclicals have already paid. Micron (MU) is the standout, with an -86.9% share cut and about $4.61B coming out after a massive +414.3% gain vs cost; that is a textbook harvest of a fully priced AI-memory story.

Nvidia (NVDA) is next, down -36.7% with roughly $1.37B in capital redeployed despite still being up 38.5% vs cost. Citadel is not abandoning GPUs; it is capping single-name risk and spreading the bet across semis and tools. Broadcom (AVGO) and Applied Materials (AMAT) are also trimmed modestly, another sign of locking in substantial gains.

Outside pure tech, they are humble about where their thesis looks less compelling. Tesla (TSLA) is cut -41.4% and Norfolk Southern (NSC) -39.5%, both with negative or middling economics relative to stronger opportunities elsewhere. Meta (META), down -38.8% and underwater at -11.7% vs cost, looks like a risk-budget casualty in favor of cleaner, higher-conviction AI and cloud exposure.

Under the sector hood, Q2 is about transforming one dominant tech engine into three: diversified tech, health care, and defensives. Tech’s top-50 weight drops from 54.11% to 32.34%, even as they add aggressively to names like KLAC, STX, TXN, Lam Research (LRCX) and Arista — the cut comes from slashing an overgrown Micron and paring high-beta leaders like Nvidia and Meta.

Health care jumps from 1.61% to 5.16% on the back of LLY, ABBV and Abbott (ABT), which together now represent a genuine second leg of earnings power. That’s a clear statement: GLP-1s, immunology, medtech and diabetes are no longer a trade; they’re a structural theme in this book.

The third leg is a blend of staples, energy and quality financials. Consumer staples rise from 3.19% to 4.23% via Keurig Dr Pepper (KDP), Mondelez (MDLZ), Coca-Cola (KO) and a still-sized CVS. Energy doubles from 1.14% to 2.16% through Chevron (CVX) and Exxon (XOM), while finance creeps up to 5.31% with new capital into BlackRock (BLK), Wells Fargo (WFC) and S&P Global (SPGI). All of this is wrapped in the 26.2% “unclassified” bucket where IVV and Berkshire Hathaway (BRK.B) sit — effectively turning a hyper-tech book into a market-plus, theme-satellite construct.

Taken together, this quarter says Citadel wants to stay long the AI and U.S. growth story, but with less path dependency and headline risk. Parking 7.6% of the book in IVV while still running concentrated AI plumbing, platform, and pharma bets gives them room to be wrong on individual names without blowing up the aggregate P&L.

Expect the AI theme to keep migrating from obvious GPU/memory winners into more infrastructure-heavy and bandwidth-sensitive stories — more Lam, KLA, Arista, storage, and data platforms, funded by any further over-earning in the old leaders. The build-out in pharma, consumer staples and integrated oils suggests a willingness to own cash-flow compounders that work under higher-for-longer rates and in choppier tapes.

The message is that Citadel is not derisking from markets; it is derisking from narratives. Q2 2026 positions the book to keep monetizing AI and U.S. consumption while being far less hostage to whether a handful of mega-cap tech tickers keep surprising on the upside.