Invesco Ltd 13F Portfolio

The biggest adds tell a very specific story: Invesco wants to own every chokepoint in AI compute, not just the headline GPU winner. NVDA, AMD, AVGO, MU, and ASML all see massive capital added, signaling a view that capacity, not demand, will be the constraint to monetize.

• NVDA, AMD, and AVGO: Sharing the load at the heart of AI acceleration, Invesco adds heavily to all three, treating them as a semi oligopoly with durable pricing power across data centers and high-end compute.
• MU, WDC, and STX: Big builds in memory and storage show a thesis that AI workloads are as much about bandwidth and persistence as about flops — the “bits and bytes” layer is being repriced.
• LRCX, AMAT, KLAC, ASML, and ARM: The step-up in semi-cap and design IP is a call that the capex super-cycle into advanced nodes and packaging is still early.
• MSFT, GOOGL/GOOG, AMZN, and META: Invesco is paying up to scale its exposure to the hyperscale platforms that actually capture AI software economics, not just sell infrastructure to others.
• PANW, FTNT, CRWD, PLTR, DDOG, SHOP, APP: Security, observability, data platforms, and software-native commerce are the application layer bet — that AI and cloud-native architectures drive higher ARPU and stickier enterprise budgets.

The twist: they are willing to add to certain names that are down versus cost (CRWD, APP, PEP, ISRG), suggesting they view recent drawdowns as volatility inside a still-intact long runway, not broken theses.

Only one name shows up as a notable trim, but it’s revealing. Caterpillar is down -14.7% in shares after a strong gain versus cost, even as AI infrastructure exposure is being scaled aggressively.

• CAT: With gain_vs_avg_buy_pct near 194%, the fund is clearly harvesting a successful industrials trade. Reducing CAT looks like a funding decision: free capital from a late-cycle machinery winner to feed earlier-stage earnings ramps in semis and AI.

The absence of other big trims in the top 50 does not mean there were no exits elsewhere, but within this visible book the message is that Invesco is not rotating within tech — it is rotating into tech from elsewhere. Profits from more cyclical or rate-sensitive exposure (as hinted by shrinking overall Finance and Real Estate weights) are being redeployed into the AI, cloud, and digital consumption complex.

For existing compounders like LLY, JPM, JNJ, and V, the pattern is small, incremental adds rather than sells. This signals ongoing respect for their compounding machines, but they are clearly no longer the primary source of upside for the portfolio.

The sector chart shows a subtle but telling re-leveraging into technology. Tech rises from 72.82% to 74.16% of the top-50 book — a large move in dollar terms given Invesco’s scale — while Consumer Discretionary also edges higher.

On the other side, Finance drops sharply from 1.46% to 0.58%, and Real Estate (misclassified Visa) falls from 1.17% to 0.48%. That’s classic "sell the balance-sheet plays, buy the earnings-duration names" behavior in a market that is rewarding secular growth over rate sensitivity.

Health Care steps down from 5.16% to 3.95%, but the remaining sleeve is more concentrated in high-ROIC, innovation-driven names like LLY, VRTX, ISRG, AMGN, and GILD. Industrials also edge lower, driven by the CAT trim even as TSLA is aggressively added as a software-and-autonomy call option inside an auto shell.

Consumer Discretionary growth — via AMZN, COST, WMT, NFLX, BKNG, SBUX, and MAR — shows Invesco prefers scale platforms and experiences where AI and data analytics can magnify margins, rather than pure-play cyclicals. The net effect is a portfolio that is more exposed to secular, data-driven business models and less tethered to credit spreads or commodity cycles.

Taken together, this quarter’s moves say Invesco believes we are still early in the AI monetization curve. The portfolio is constructed on the premise that earnings power will accrue in layers: from foundry tools and memory, to accelerators, to hyperscale cloud, to security and data platforms, and finally to consumer and enterprise applications.

The heavy adds to NVDA, MU, AMD, AVGO, LRCX, AMAT, KLAC, ASML, and ARM suggest they see the capex super-cycle and memory upturn as multi-year — not quarters-long — phenomena. Meanwhile, scaling positions in MSFT, GOOGL, GOOG, AMZN, META, and SHOP indicates a belief that AI will translate into measurable pricing, mix, and share gains at the platform layer.

By cutting back on CAT and letting Finance and Real Estate shrink as slices of the book, they are explicitly sacrificing some cyclical diversification for more exposure to long-duration growth and intangible-asset moats. The remaining Health Care and Consumer Staples names (LLY, VRTX, JNJ, AMGN, GILD, PEP) look like ballast with upside, not primary risk drivers.

If this playbook holds, expect Invesco’s future quarters to be judged less by sector rotation and more by intra-stack adjustments — shifting among GPUs, memory, tools, and AI software — as they refine where in the value chain the incremental dollar of AI profit will ultimately land.