California Public Employees Retirement System 13F Portfolio

The biggest single statement this quarter is the $10.4B lift in IVV, up +1097.6% in shares to 6.41% of the book. That is not a tweak — it is CalPERS explicitly choosing S&P 500 beta as a primary risk unit alongside VOO, which still sits at 11.89% despite being trimmed. They are taking some stock‑specific AI risk off and replacing it with index risk.

Under the surface, the incremental dollars are telling you they still believe in AI, but via the picks‑and‑shovels layer rather than just the poster children:

• IVV: A massive add that reaffirms a core bet on U.S. large‑cap equities as the default destination for harvested AI gains.
• AMAT, AMD: Both increased (AMAT up +15.5% in shares, AMD up +11.9%), signaling confidence in continued capex for leading‑edge fabs and compute, even as they cut back NVDA and LRCX.
• STX, SNDK, WDC: Aggressive adds to storage and components (STX shares up +93.7%, SNDK +33.7%, WDC +14.3%) show a thesis that AI’s data footprint will keep driving demand for capacity and high‑value storage hardware.
• TSLA, CAT, GEV: Increases across autos and heavy equipment (TSLA +7.6%, CAT +9.7%, GEV +18.2%) suggest a complementary bet on industrial demand and electrification, funded by gains elsewhere in tech.
• C and the rest of the bank complex: Citigroup’s shares are up +38.3%, with smaller increases in WFC and modest tweaks elsewhere, pointing to an opportunistic, valuation‑driven add in large banks as rates and credit conditions normalize.

The through‑line: they’re still structurally long the AI build‑out, but the marginal dollar is now going into the infrastructure and the index, not just the headline platforms.

The funding side of the ledger is just as deliberate. CalPERS is taking real money off the table in the very names that powered its last three years of outperformance, but it’s scaling, not abandoning, those bets.

At the top of the trim list sit the AI and megacap growth icons:

• NVDA: Shares cut by -32.6%, freeing up about $4.45B, even though the position still sits at 5.2% of the book and roughly +276.8% above their average cost. This is classic risk management after a parabolic move.
• AAPL, MSFT: Apple (-21.4% in shares) and Microsoft (-27.9%) are both long‑held and deeply in the money; trims here recycle capital from crowded, expensive compounders into broader beta and cheaper cyclicals.
• GOOGL, GOOG, META, MU, LRCX, MRVL: Across Alphabet’s share classes, Meta, and a swath of memory and equipment names, they’re shaving 15–35% of share count — locking in huge gains (Micron is more than +500% vs cost) while keeping meaningful exposure.

Outside of tech, the behavior rhymes. Large, profitable but slow‑growth pharma (LLY, JNJ, MRK, AMGN, ABBV) are all reduced, alongside staples and telecom (KO, CSCO, VZ). These are not thesis reversals; they are liquidity pools, drained to fund higher‑conviction growth in AI infrastructure, industrials, and the index core.

Even VOO gets tapped: an -8.7% reduction in shares frees about $2.01B that effectively migrates into IVV and the targeted adds described earlier. For a giant plan like CalPERS, the pattern is clear: sell what worked too well, keep a sizable stub, and move the marginal dollar where the multi‑year risk‑reward looks better.

CalPERS is not “out of tech” — at 45.16% of the top‑50 book, technology is still the anchor — but the composition of that tech risk is migrating from mega‑cap platforms toward diversified exposure and hardware enablers. The raw tech weight dropped from 52.8% to 45.16%, driven almost entirely by profit‑taking in the largest AI and cloud names.

The largest net gainer on the sector chart is the catch‑all “Unclassified” bucket, which jumps from 22.46% to 30.79%. That’s primarily the surge in IVV alongside a still‑enormous VOO position and a steady JPHY corporate bond ETF weight, plus BRK.B and GEV. In practice, a big chunk of the fund’s tech, industrial, and financial exposure is now held via these wrappers rather than individual names.

Elsewhere, the shifts are incremental but telling:

• Health care ticks down from 5.78% to 4.98% as they lighten up on big‑pharma winners after a monster run.
• Finance edges up from 3.97% to 4.08% on higher C and WFC exposure, indicating measured confidence in large banks.
• Industrials move from 2.5% to 2.85% on added TSLA and CAT, and Energy inches up (1.32% to 1.42%) via modest XOM and CVX adds.
• Telecommunications shrinks from 1.63% to 1.14% after sizable cuts in CSCO and VZ, underscoring that legacy telco is a funding source, not a growth theme.

Net‑net, they’re still heavily growth‑biased, but more of that growth is held in indices and cycle‑sensitive enablers, with slightly less tied to the most sentiment‑driven AI beneficiaries.

Viewed in context of a 3‑year annualized return of 25.4% and a 5‑year run over 15%, this quarter looks like CalPERS converting a trader’s AI windfall into a durable, institutional portfolio. The AI bet persists, but the risk is being spread: less concentration in a handful of names, more in the supply chain and in the S&P 500 itself.

Expect more of the same if markets stay strong. As long as AI‑linked names like NVDA, MU, and the storage complex remain well above cost, they’re natural sources of cash to feed incremental adds in banks, industrials, and broad beta ETFs. If volatility spikes, the enlarged IVV and VOO sleeves give them cheap, liquid levers to dial risk up or down without touching individual positions.

Crucially, the modest but broad‑based adds in cyclicals and financials suggest they see more runway in the real‑economy leg of this cycle. TSLA, CAT, GEV, XOM, CVX, and the banks build a narrative of a plan that wants exposure not just to digital infrastructure, but also to physical and financial infrastructure as AI diffuses into the broader economy.

For outside observers, the message is straightforward: CalPERS is not making a dramatic style call; it is professionalizing an AI‑heavy growth portfolio into something that can be run at pension scale. Less headline risk, more beta, and a deeper bench of enablers — that’s the 2026‑Q2 signature.