Rhumbline Advisers 13F Portfolio

The biggest buys table reads like a shopping list for the less glamorous, more torque‑y parts of the AI supply chain.

• Marvell Technology (MRVL) is the standout: shares are up +64.1% with an estimated +$174.5M added, a decisive vote that custom accelerators and networking silicon will catch the next leg of AI capex rather than just riding NVIDIA’s coattails.
• Micron (MU) saw a +3.3% share add worth roughly +$72.5M; that’s a direct bet that high‑bandwidth memory and DRAM pricing will remain tight as model sizes and training intensity keep climbing.
• Alphabet’s two share classes (GOOGL and GOOG) were both increased, by +1.4% and +1.0% respectively, adding about +$52.0M and +$27.9M; that reinforces the thesis that hyperscale AI platform economics (search, cloud, ads) still compound even if hardware cycles whipsaw.
• Western Digital (WDC) got a +6.5% bump (~+$24.8M), and Applied Materials (AMAT) and AMD were also increased, extending the bet further down into storage and equipment as well as an alternative GPU/CPU vendor.
• Even a cyclical industrial like Caterpillar (CAT), up +2.1% (~+$18.0M added), fits the pattern: a modest nod to real‑asset capex and infrastructure demand that AI‑driven productivity and reshoring may stimulate.

Net‑net, the manager is deliberately sliding incremental capital from the AI front men to the backline — the companies that sell the bandwidth, bits, and boxes required to keep the story running.

On the sell side, the pattern is clear: Rhumbline is clipping exposure in the richest, most consensus trades to fund higher‑beta expressions of the same themes.

• Apple, Microsoft, Amazon, and NVIDIA are all tagged as decreased, with Apple down -2.9% (about -$214.4M), Microsoft -3.0% (-$144.2M), Amazon -3.4% (-$134.4M), and NVIDIA -1.1% (~-$92.0M); that looks like disciplined profit‑taking in positions showing enormous gains versus average cost.
• Outside the core AI platforms, they’re shaving health‑care winners like AbbVie (-5.7%, roughly -$41.4M) and Eli Lilly (-1.9%, about -$30.5M), plus large banks such as JPMorgan (-2.2%, -$33.4M), converting defensive and rate‑sensitive winners into ammo for semis.
• Costco, Coke, and other staples/retail names are also gently trimmed, reinforcing the idea that the portfolio’s return engine is expected to come from tech and AI, not from consumer defensives.

Importantly, these are single‑digit percentage reductions, not wholesale repudiations. The manager still wants the earnings compounding and cash‑return profile of megacap tech and big pharma in the book; they just no longer want all of their AI upside tied to the four or five most crowded tickers in the market.

At the sector level, this quarter looks almost static: technology edges from 63.28% to 63.6%, while consumer discretionary, health care, finance, and energy all tick down only a few basis points. On paper, it’s a steady‑state allocation.

Under the hood, though, tech is quietly reshaping. Within semiconductors, capital is migrating from the dominant GPU and CPU franchises toward memory (MU), storage (WDC), and diversified chip names like MRVL and AMAT that are more leveraged to AI infrastructure spend than to headline unit volumes.

Industrials inch up from 4.13% to 4.22% on adds to TSLA and CAT, a small but telling move that pairs “future mobility” with old‑line machinery as parallel bets on a capex‑heavy world. Health care (6.86%) and finance (5.03%) drift slightly lower as winners are trimmed on strength rather than abandoned.

Real estate (in practice, Visa and Mastercard) and energy remain modest sleeves, effectively serving as balance‑sheet quality and cash‑flow ballast rather than areas where the manager is trying to generate alpha this quarter.

Putting it together, Rhumbline looks committed to an AI‑centric equity regime, but is deliberately broadening its exposure from a narrow band of platform megacaps into the messier, more cyclical parts of the hardware stack. That shift should increase sensitivity to capex and pricing cycles in memory and storage, but also introduces more upside torque if AI infrastructure demand keeps surprising on the upside.

The light trims in banks, energy, and consumer defensives suggest the manager isn’t positioning for an imminent recession; instead, they’re accepting some macro noise in exchange for owning the dominant franchises in data, compute, and payments. Small adds to UNH and MRK show they still value steady cash‑flow compounders as a hedge against tech multiple risk.

Going forward, watch whether MRVL, MU, WDC, and AMAT keep climbing the rankings at the expense of Apple, Microsoft, and Amazon. If that continues, it will confirm that Rhumbline wants to be paid not just for AI existence, but for AI’s appetite for bits, bandwidth, and bulldozers.