Citigroup INC Doubles Down on AI Platforms and Cyclical Rebound in 2026-Q1

The biggest dollar adds are brutally consistent: concentrate in AI, then surround it with beneficiaries and cyclical torque. Nvidia, Microsoft, Apple, Alphabet (both GOOGL and GOOG) and Broadcom remain the core bet that training and inference economics will keep compounding for years.

Below the obvious mega-caps, they quietly upgrade the plumbing of the AI stack. Micron, AMD, TSMC, Applied Materials, Lam Research, KLA, Texas Instruments, Qualcomm, and Cisco all see position increases, signaling a belief that memory, analog, foundry capacity, networking, and tools will capture the second derivative of AI demand.

Outside pure tech, conviction is rising in three pockets:

• IWM and QQQ: a sharp ramp in small caps and a bigger slug of Nasdaq growth instead of owning the S&P wrapper.
• Banks: Bank of America, JPMorgan, and especially Deutsche Bank (+257.4% shares) show a bet on credit normalization and steeper curves.
• Health care: Eli Lilly (+65.5%) and a new AstraZeneca stake (about $717.8M) elevate GLP‑1s and oncology into core growth, while Johnson & Johnson is scaled up as a diversified cash‑flow compounder.

They also move aggressively into software beta via IGV, with a +690.2% share jump, effectively creating a mid‑cap cloud and SaaS basket around the mega-cap AI names. Add Caterpillar’s +134.0% share increase and more XLE/XOM, and you have a clear reflation sleeve riding infrastructure, energy, and industrial capex.

Funding for this repositioning comes from one loud source: de‑risking the bland middle of the portfolio. Slashing SPY by 80.8% and materially cutting TLT says they see little payoff in owning the entire market or long duration when they have well-defined winners to back.

The TLT cut is especially telling; at a loss relative to their average cost (negative gain_vs_avg_buy_pct), it looks less like profit‑taking and more like a macro call that bond upside is capped relative to equities and credit. They want risk, just not in Treasuries.

Equity trims are pointed, not panicked:

• Tesla is down 20.5% in shares, even though they are well ahead of cost, hinting at recycled capital from speculative growth into more predictable AI and health care earnings.
• Intel is reduced by 21.0%, despite a strong gain vs cost, suggesting they prefer higher‑quality semi exposure in Nvidia, TSMC, and the tools vendors.
• XBI and IBB are both nudged lower, freeing capital from broad biotech beta to fund focused exposure in Lilly, AstraZeneca, and Johnson & Johnson.

Goldman Sachs is cut by 23.5% while Bank of America, JPMorgan, and Deutsche Bank are increased, a clear tilt away from fee‑driven investment banking toward deposit‑funded and rate‑sensitive balance sheets. VOO and IAU are only marginally trimmed, which looks more like housekeeping than a real view change.

Sector-wise, this is a decisive rotation into concentrated themes. Technology rises from 50.16% to 54.26%, not by adding fringe names but by doubling down on the hardware, software, and infrastructure feeding AI spend.

The “Unclassified” bucket falls from 27.87% to 21.41% as they unwind broad wrappers (SPY, some VOO) and long bonds (TLT), while upgrading specific ETFs that express sharper views: IWM and KRE for small and regional banks, IGV for software, XLE for energy, XLI for industrials, and GLD for gold.

Health care jumps from 1.82% to 3.28%, and the character of that exposure changes. It shifts from ETF-heavy, early-stage biotech risk (now partially reduced via XBI/IBB trims) toward large-cap pharma with direct earnings from obesity and oncology pipelines.

Finance climbs from 4.71% to 5.76%, but the mix is striking: more money‑center banks and European credit (DB in particular), less Goldman. Industrials edge down from 4.42% to 3.75% on the Tesla trim but are refocused into Caterpillar’s hard-asset leverage to infrastructure and energy. Energy itself is roughly stable in headline weight, but with a clear tilt to owning the sector outright through XLE and scaled XOM exposure rather than just oil majors as bond proxies.

Pull the quarter together and the picture is sharp: Citigroup wants to be over‑exposed where earnings growth is durable and under‑exposed where valuations are just beta. A 24.0% top‑10 concentration anchored by Nvidia, Microsoft, Apple, Amazon, and Alphabet means their performance will live or die with AI adoption and cloud‑driven workloads.

The reflation basket – small caps, regional banks, Caterpillar, XLE, and XOM – says they see more upside in the real economy if growth and nominal GDP stay firm. Cutting TLT and SPY to fund those adds reinforces that they are not hiding from macro volatility; they are leaning into it where risk‑reward looks asymmetrical.

Health care’s upgrade and the choice of Lilly, AstraZeneca, and Johnson & Johnson position the book to capture multi‑year drug categories with pricing power, which can work in both inflationary and slow‑growth environments. Meanwhile, a higher GLD stake alongside modest IAU preserves a tail hedge against policy error and geopolitical shocks.

In short, they are building a barbell: AI platforms and software on one side, tangible‑asset cyclicals and large pharmas on the other, with less generic beta in between. If AI capex, infrastructure spending, and obesity/oncology drugs do what the market narrative currently promises, this portfolio is set up to outperform broad indexes — and it is now structurally less interested in matching them.