Primecap Management CO Ca Rotates From AI Mania To Defensible Cash Flows

The clearest statement of rising conviction is the creation of a new, high‑weight anchor in global pharma alongside an already massive Eli Lilly stake. Health care’s weight gain is not a drift; it is the product of deliberate, multi‑billion‑dollar allocation.

• AZN (AstraZeneca) arrives as a brand‑new position at 2.83% of the book, worth $3.59B. That instantly makes it one of the largest holdings, effectively building a second big‑pharma pillar opposite Lilly’s 6.68% and signaling belief that late‑stage pipelines and specialty oncology are a better use of marginal dollars than chasing ever‑richer AI multiples.
• BSX (Boston Scientific) is quietly but meaningfully increased, up 4.6% in shares and about $89.9M. Upping a medical device compounder that already sits at 1.61% suggests a preference for procedure‑driven volume growth and recurring device revenue as a complement to binary biotech risk.
• ADBE (Adobe) is the standout software add, with shares up 28.7% and an estimated $256.4M of fresh capital. After a bruising reset in high‑multiple software, choosing to scale Adobe — rather than broad software beta — looks like a targeted bet on entrenched creative and marketing moats that still benefit from AI, but on subscription‑like rails.
• SCHW (Charles Schwab), V (Visa), IEX (IDEX) and BMY (Bristol‑Myers Squibb) all see incremental adds, each only around $6–$15M but directionally consistent: more transaction rails, more fee‑based financials, more industrial components, and more value‑tilted pharma exposure.

Taken together, the big buys say Primecap wants its incremental risk in businesses with identifiable cash engines and long product cycles, not in the most crowded parts of the AI narrative.

On the sell side, Primecap’s message is subtle but clear: take chips off the table where AI gains are front‑loaded, keep core exposure, and fund upgrades elsewhere. The biggest dollar reductions cluster in semis and tech hardware, plus a few mature financials and cyclicals.

• MU (Micron) is the largest trim by far, with shares down 11.8% and roughly $986.7M freed up. That is not a thesis abandonment — it still stands as a 5.82% position — but the size of the harvest underscores how much of the near‑term AI memory trade they view as already captured.
• KLAC (KLA) gets cut more aggressively on a percentage basis, with a 15.2% share reduction and about $645.4M pulled out. When process‑control capex beneficiaries are trimmed alongside Micron, it looks like a broad decision to reduce sensitivity to the next leg of the semiconductor capex cycle.
• FLEX (Flex), TXN (Texas Instruments) and ORCL (Oracle) are also clipped, yielding $52–$86M each. These were big historical winners with four‑digit or triple‑digit gains versus cost; trimming them looks like a valuation discipline exercise rather than a fundamental capitulation.
• Outside tech, FDX (FedEx), TSLA (Tesla), JPM (JPMorgan), WFC (Wells Fargo) and several travel/consumer names see modest position shrinkage. Primecap is keeping the cyclical call — freight, autos, cruise lines, airlines, off‑price retail — but clearly using them as liquidity sources to finance higher‑conviction core holdings.

The pattern across sells is consistent: lower the beta and capital intensity at the margin, harvest outsized gains, and redirect to health care, software, and fee‑driven financials.

The sector bars tell a simple story: a high‑conviction tech investor rebalancing toward a sturdier health care spine. Technology falls from an estimated 39.38% to 37.0%, while health care jumps from 25.16% to 28.53%.

That shift is almost entirely explained by the new AstraZeneca stake layered on top of Lilly, Amgen, Biogen and a host of other pharma/biotech names. Health care now rivals technology as the core of the book, giving the portfolio a second growth engine with very different cycle drivers — demographics and innovation rather than capex and cloud budgets.

Consumer discretionary edges down slightly to 16.01% from 16.46%, but remains a meaningful lever via FedEx, Amazon, off‑price retailers and travel/leisure, all of which keep the portfolio exposed to the health of the US consumer and corporate spend. Finance dips to 6.34% from 6.55%, yet the internal rotation is telling: Schwab and Raymond James grow, while large banks like JPMorgan and Wells Fargo are trimmed, nudging the book toward capital‑light fee income over balance‑sheet leverage.

Industrials, energy, and a small consumer‑staples sleeve anchored by Sony stay roughly stable, preserving diversification but clearly playing supporting roles. The net result: less dependence on the next leg of the AI trade, more ballast from recurring health care and transaction economics.

Read through the quarter, Primecap is signaling confidence that the AI and digital infrastructure themes it backed early are now self‑funding — and that the better place for fresh risk is in compounding cash flows with longer, smoother runways. The health care build‑out, especially AstraZeneca and added medtech, positions the portfolio to benefit from multi‑year drug and device adoption curves that are relatively insensitive to short‑term macro swings.

On the tech side, the pivot from trimming Micron and KLA toward adding Adobe suggests a desire to own the application and workflow layers of AI as much as the hardware. If semis remain volatile, software and payments names like Adobe and Visa can still capture AI‑driven productivity and transaction growth with less capital intensity.

The modest de‑risking in banks and high‑beta cyclicals, without wholesale exits, indicates Primecap still wants exposure to a benign credit and consumer backdrop — just not as the primary driver of returns. Instead, it is concentrating incremental dollars in businesses with durable moats, high switching costs, and pricing power.

Going forward, expect the book to remain tech‑tilted but increasingly defined by a barbell: AI‑levered infrastructure and software on one side, and an expanding cluster of large‑cap pharma and medtech on the other. If that balance is right, the portfolio should be less about calling the next AI leg and more about letting structurally advantaged franchises grind out excess returns over time.