Capital Research Global Investors Doubles Down on AI Platforms and Risk-On Cyclicals

The biggest dollar adds cluster tightly around a single idea: own the platforms that monetize AI and the networks through which global activity flows. When your top incremental dollars go into Broadcom (est. +$9.18B), Microsoft (est. +$8.43B), Meta (est. +$7.46B), Amazon (est. +$5.91B), and Alphabet (est. +$3.50B via GOOG alone), you are declaring that the margin stack of the AI era will live in hyperscale hardware, cloud, and ad-driven attention.

Several of these positions are already very profitable: Broadcom is up 317.9% versus its average cost, Microsoft 128.8%, Nvidia 116.5%, Amazon 110.8%, and Alphabet’s GOOGL and GOOG lines over 180.9% and 206.4% respectively. Adding to winners at these gain levels is an explicit stance that their competitive moats are widening, not narrowing, as AI spend ramps.

The new position in Cisco, sized at 0.77% or about $5.0B, says they’re also willing to pay for the connective tissue as traffic and latency demands explode. This complements existing bets on semiconductor plumbing (Broadcom, Nvidia, Intel, TSMC, Applied Materials, Amphenol) and suggests a full-stack infrastructure thesis rather than a narrow bet on any one AI model provider.

Conviction is also quietly ramping in scale compounders beyond tech. Philip Morris is up +46.3% in shares, Linde +58.5%, Union Pacific +54.0%, and Mondelez +131.4%, indicating they want more exposure to essential, oligopolistic businesses that can push price through cycles. In financials, outsized increases in Wells Fargo (+94.8%), Progressive (+102.6%), and BlackRock (+20.1%) show a belief that credit and markets can absorb higher volatility without breaking the system.

Even some underperformers are being averaged into. Medtronic is added to despite a -17.4% mark-to-cost, and Carvana’s stake is up +90.3% with the position sitting -78.9% versus average buy. That signals selective willingness to underwrite idiosyncratic turnarounds where optionality is high relative to current mark.

The sell-side of the ledger is small but revealing. RTX is the largest trim by dollars, with shares cut -31.3% and an estimated -$4.59B taken off the table, even though the position still shows a 111.4% gain versus average cost and remains a meaningful 1.57% of the book.

Applied Materials is reduced -9.9% (about -$1.11B) despite a 110.1% gain versus cost, which looks more like risk budget reallocation within the semiconductor complex than a call that the wafer equipment cycle is over. They are clearly happier to pile into Broadcom, Nvidia, Intel (+190.4% shares), and TSMC (+210.7% shares) than to add further to the fab tools vendor at this stage.

Dominion is trimmed by -18.1% (roughly -$0.67B), even as it remains in the money by 13.4%. With utilities overall ticking up slightly in weight thanks to a 245.1% surge in CenterPoint, this is a portfolio geometry move: keep some regulated yield exposure but rotate from a more fully valued incumbent into a name with more perceived upside.

Importantly, there are no large, directional evacuations from any core theme in the visible top-50. Industrials overall are down, but RTX remains a top-20 line, and TransDigm and Union Pacific are both being increased. Energy’s slight step-down is accomplished via lack of big adds, not heavy selling of Exxon, Canadian Natural Resources, or EOG, each of which actually sees modest share increases.

On the surface, sector weights barely budge: technology sits at 45.78% of the book versus an estimated 45.76% previously. Underneath that stability, the risk mix shifts toward large, diversified platforms and away from narrower cyclicals like Applied Materials.

Health care drifts down from an estimated 16.89% to 16.14%, but the pattern is consolidation, not abandonment. They are leaning harder into big, proven franchises like Eli Lilly, Vertex, AbbVie, Abbott, and UnitedHealth (shares up a striking +173.3%), while keeping optionality in smaller innovators like Alnylam and devices via Medtronic despite drawdowns.

Consumer is where the risk dial clearly turns up. Consumer discretionary rises to 12.6% from 11.63%, with builds in Amazon, Starbucks, Royal Caribbean, Netflix, Home Depot, and Carvana. Consumer staples also lift to 0.88% from 0.48%, driven by a sizable Mondelez add, giving them some ballast alongside the more volatile leisure and e‑commerce plays.

Industrials step down to 6.16% from 7.82%, a function of the RTX trim and prior-size choices, even as Carrier (+55.7%), Union Pacific (+54.0%), TransDigm (+26.9%), and Tesla (+5.1%) are all increased. Energy eases from 4.65% to 3.90% but with small positive flows into Canadian Natural Resources, Exxon, and EOG, implying no strong macro call against hydrocarbons.

Finance edges up from 5.08% to 5.82% as banks, insurers, and asset managers are all topped up. Real estate-type exposures (Uber, Mastercard, MercadoLibre in this classification) drift down a bit, while basic materials (Linde), utilities (via CenterPoint’s big build), and telecom infrastructure (new Cisco at 1.11% sector weight) are being used as durable, cash-generative anchors around the higher-beta AI and consumer bets.

Put together, this quarter’s moves sketch a barbell: on one side, a concentrated bet on a handful of global AI and cloud oligopolies; on the other, a diversified set of cash-generative cyclicals and defensives that can survive a choppier macro tape. The fund is effectively saying that the next leg of equity returns will be driven by compute intensity and scale advantages, not by rate compression or one-off cost cutting.

Doubling down on winners like Broadcom, Microsoft, Nvidia, Amazon, Meta, and Alphabet at triple-digit percentage gains versus cost is the core tell. They are comfortable owning the perceived “over-earning” names because they view the earnings base as structurally underappreciated relative to the AI demand curve.

At the same time, they’re rebuilding cyclical and financial shock absorbers: banks (JPMorgan, Wells Fargo), capital markets (Morgan Stanley, BlackRock), insurance (Progressive), rails (Union Pacific), aerospace (RTX, TransDigm), and industrials (Carrier, Tesla) all feature as scaled beneficiaries of a world where nominal GDP and trade volumes stay healthy. Consumer exposure is biased to franchises that monetize time and experiences — from Starbucks and Royal Caribbean to Netflix and Home Depot.

Health care, utilities, and staples form a stabilizing sleeve rather than a primary source of alpha. The sizable increases in UnitedHealth, Abbott, AbbVie, Eli Lilly, Linde, CenterPoint, and Mondelez suggest a desire to keep a floor under portfolio volatility without sacrificing pricing power.

For observers, the implication is straightforward: unless the AI capex cycle abruptly stalls or a deep recession hits both consumer and credit, this portfolio is built to ride an extended phase of higher-for-longer nominal growth. Capital Research Global Investors is not trying to time a peak; it is trying to own the infrastructure, platforms, and brands that will keep compounding as that peak keeps moving out.