Aqr Capital Management 13F Portfolio · Cliff Asness

The biggest dollar add this quarter was not some obscure small cap; it was yet more NVIDIA. AQR lifted its stake by +17.8% to $7.48B, even though the position is already up 98.9% versus its blended cost. That is a deliberate choice to press the AI platform leaders rather than declare victory.

The same pattern shows up across the other hyperscale beneficiaries:

• Apple: shares up +20.3%, value to $5.01B, with a 94.9% gain vs average cost.
• Microsoft: +14.3% shares, now $4.26B, still adding despite a 55.2% gain.
• Alphabet (GOOG): +21.8% shares, $2.24B, nearly doubling their money vs entry.
• Amazon: +25.9% shares, $2.84B, as the cloud/commerce flywheel keeps paying.

Under the surface, AQR is also upgrading its AI infrastructure footprint. It added to Broadcom, Lam Research, Fortinet, Seagate, Vertiv, and AMD — a cluster that ties directly into networking, wafers, security, and power/thermal kit around data centers. Carpenter Technology, doubled to $900.2M, fits the same story from the specialty metals side.

On the macro side of the book, the biggest percentage and dollar adds were to money‑center banks. Wells Fargo was boosted +274.0% to $888.4M, JPMorgan +74.4% to $930.3M, and Capital One +45.9% to $1.06B. Chevron’s stake ballooned +268.6% to $912.1M, giving the fund a more explicit energy and rate‑sensitive tilt than in prior quarters.

Even the more defensive adds, like Verisign and Travelers, sit at the intersection of recurring cash flows and pricing power. The message is coherent: AQR isn’t chasing speculative AI; it is concentrating capital in scaled, cash‑rich beneficiaries and the capital‑intensive infrastructure those winners require.

If the buys say “own the AI platforms,” the sells say “harvest the peripherals.” Micron remains a large $4.43B position, but AQR cut shares by -15.2% after a 476.7% gain versus cost. Western Digital and legacy SanDisk exposure were also reduced by -15.0% and -35.8% respectively, with SanDisk up an extraordinary 2795.2% on their basis.

Those aren’t risk‑off trades; they’re classic rebalancing from the noisiest AI cyclicals into stickier franchises. Arista Networks, central to high‑speed networking, was cut -25.4% despite more than doubling from cost. Booking Holdings, another structural grower, saw a -36.0% trim, likely to free capital for Amazon, Airbnb, and Costco — all of which were increased.

Funding also came from the defensive side of the ledger. In health care, AQR trimmed Centene (-14.7%), Bristol‑Myers (-7.7%), and Merck (-5.7%), even as it added to higher‑octane Eli Lilly. Utilities were lightened as well: Edison International was reduced -10.3%, while PG&E was left roughly unchanged. TechnipFMC, a more equipment‑heavy energy name, was cut -17.5% even as Chevron was aggressively scaled.

This is not a wholesale sector exit — Technology is still 54.06% and Health Care 6.99% of the disclosed book. Instead, AQR is rotating within its long‑standing exposures, pruning the most cyclical or lower‑quality edges to augment the core platforms and balance‑sheet winners it wants to own into the next leg of the cycle.

The sector chart looks tame at first glance, but the internals matter. Technology’s weight slipped only modestly from 55.0% to 54.06%, even after big trims in Micron, Western Digital, Arista, and SanDisk. That persistence — alongside huge adds in NVIDIA, Apple, Microsoft, Alphabet, AMD, and Vertiv — shows they’re not backing away from AI; they’re reshaping it toward durable, ecosystem‑wide exposure.

The more pronounced shifts are in the macro‑sensitive parts of the book. Finance climbed from 6.25% to 7.63% as AQR piled into Wells Fargo, JPMorgan, Capital One, Travelers, and Chubb. Energy jumped from 0.33% to 1.12% on the back of the Chevron add, while industrials rose from 9.96% to 10.44% with extra capital in Tesla, Cummins, Carpenter Technology, Teradyne, JB Hunt, and GM.

On the other side, Health Care slid from 7.95% to 6.99% and Utilities from 3.98% to 3.49%. Those cuts, plus a move down in telecom/networking exposure via Arista, freed budget for the reflation basket without meaningfully changing the consumer footprint. Consumer Discretionary barely budged (11.47% to 11.48%), but the mix tilted up‑market — more Amazon, Airbnb, Costco, TJX, and Viking, and less Booking.

Net‑net, AQR is still a quant shop with a tech core, but the factor dial has shifted. The book now carries more rate and commodity beta through banks and integrated energy, and slightly less low‑vol defensiveness, than it did three months prior.

Taken together, these moves look less like a hot‑money chase and more like a multi‑year positioning for AI‑driven capex and a stickier‑inflation world. AQR is willing to own the obvious winners — NVIDIA, Apple, Microsoft, Alphabet, Amazon, Meta — at nearly double their cost because the portfolio is hedged with cyclicals that benefit if growth and rates stay firm.

The trims in memory, storage, and networking (Micron, Western Digital, SanDisk, Arista) hint that they see more volatility at the edges of the AI trade than at its core. By contrast, adds to Broadcom, Lam Research, Vertiv, Carpenter Technology, and Cummins suggest an expectation that data center, grid, and industrial investment will run longer than consensus gives credit for.

The banking and energy build‑out is equally telling. AQR turned Wells Fargo, JPMorgan, Capital One, Chevron, and the property‑casualty insurers into a more meaningful part of the book, a configuration that typically pays off if credit losses stay contained and real assets hold their pricing power. The moderate de‑risking in health care and utilities underscores that bet.

Going forward, watch whether they keep leaning into this AI‑plus‑reflation barbell. If tech multiples compress but spending on compute, networking, and grid upgrades remains robust, this mix of platforms, infrastructure, and rate‑sensitive cyclicals is set up to participate on multiple fronts — and the systematic trimming pattern suggests they’ll keep recycling gains rather than letting any one theme dominate the risk budget.