Capital International Investors 13F Portfolio

The biggest buys table reads like a shopping list for the next decade of compute buildout and hard-asset scarcity. This is not a speculative AI pivot; it’s a bet on the capital stock that makes AI (and everything else) possible.

On the semiconductor side, conviction is unmistakable:

• Intel (INTC) was taken up +437.1%, adding about $4.38B, turning it into a core AI-capacity and US-foundry reshoring bet at 1.11% of the book.
• Deere (DE) jumped +276.3% (+$2.99B) to 0.84%, signaling confidence that automation and precision ag remain under-owned industrial AI beneficiaries.
• Cisco (CSCO) appears as a new $2.66B position at 0.55%, a clear move into network infrastructure as AI traffic and security needs explode.
• KLA (KLAC) rose +30.3% with roughly $2.49B added, cementing process control as a structural bottleneck they want to own through the cycle.
• Micron (MU) was boosted +16.3% (+$1.68B), leaning into high-bandwidth memory as a key constraint for AI training and inference.
• Amphenol (APH) and ASML (ASML) saw +21.8% and +38.0% adds respectively (roughly $1.36B and $1.32B), spreading the bet across interconnects and irreplaceable EUV tooling.
• Public Storage (PSA) climbed +49.3% (+$998.1M), an old-school way to add real-asset duration and inflation protection alongside data and chips.

Across these moves, they’re upgrading from AI narrative to AI capital intensity: shovels, picks, and the grid that powers them.

The funding list for these adds is stark: they’re selling success where expectations look fattest. The largest trims by dollars are a who’s-who of mega-cap AI and aerospace winners.

• Alphabet’s GOOGL and GOOG lines were cut -20.5% and -8.3%, pulling about $1.87B and $1.34B, even as both sit hundreds of percent above their average costs. That’s profit-taking on a dominant franchise that’s become consensus AI exposure.
• Nvidia (NVDA) was trimmed -4.8% (about -$946.5M), Broadcom (AVGO) -2.6% (around -$840.0M). Both remain huge positions, but they’re clearly being rebalanced from AI GPUs and custom silicon beneficiaries to the foundry and equipment names that must get paid regardless of who wins the model wars.
• In defense/aerospace, Northrop Grumman (NOC) and TransDigm (TDG) were cut -17.5% and -7.9%, releasing roughly $633.0M and $575.4M, respectively. After big runs and elevated multiples, these look like classic sources of cash.
• Healthcare leadership names UnitedHealth (UNH) and Eli Lilly (LLY) also saw meaningful trims (-13.5% and -6.2%, around -$504.0M and -$458.2M), hinting that the fund sees better risk/reward in less-loved pharma and cashflow REITs than in the market’s highest-multiple growth defensives.

None of these exits read as a change of mind on the underlying businesses; they read as “own less of what everyone already loves, and more of what AI needs but the multiples still underprice.”

On the surface, sector allocation looks stable: Technology inches from 53.72% to 53.8%, Finance holds around 5.3%, and Industrials edge up from 7.99% to 8.21%. Underneath that, the rotation is sharper.

Within tech, they are quietly edging away from pure megacap platforms toward the supply chain: heavy adds in Intel, ASML, KLA, Micron, Amphenol, and a fresh Cisco position, funded by trims in Nvidia, Broadcom, Microsoft, and Alphabet. The net effect is more exposure to the volume of wafers, bits, and packets, and slightly less to ad budgets and consumer devices.

Outside tech, the fund is adding ballast. Real estate moves from 8.42% to 8.84%, with Welltower and Public Storage growing, reinforcing a preference for long-duration, asset-backed cashflows. Utilities (via Constellation Energy) and Industrials (Deere, Danaher, FTAI) tick higher, while Health Care, Energy, and Consumer Discretionary ease back as they trim leaders like Lilly, UnitedHealth, TotalEnergies, and travel/consumer names. The single biggest “new” sector line is Telecommunications at 0.81%, entirely driven by Cisco — effectively a call on network infrastructure under an outdated label.

Taken together, this quarter suggests Capital International Investors wants to own the AI and industrial boom in ways that are hard to dislodge and less reflexively crowded. The book is being nudged away from brand-name AI proxies and toward the inputs and infrastructure that all winners must rent.

Expect them to keep recycling gains from megacap platforms into capacity and tools: more foundry, more equipment, more networking, more interconnects. Names like Intel, ASML, KLA, Micron, Amphenol, Cisco, and Deere now form a coherent theme around capital-intensive bottlenecks, while the build in banks, card networks, utilities, and storage REITs shows a parallel desire for steady, regulated, or oligopolistic cash streams.

If this pattern continues, future quarters should show incremental trimming of the highest-multiple, most-indexed winners in favor of underappreciated enablers and asset-heavy franchises. For outside observers, the signal is clear: they see the AI and reshoring cycle less as a story about a handful of platforms, and more as a long grind of capex, supply constraints, and the cashflows that accrue to the plumbing.