Goldman Sachs Group Inc Doubles Down on AI Platforms and EM Beta

The biggest adds by dollars read like a who’s‑who of AI infrastructure and data monopolies. Microsoft (+19.1% shares, +$3.63B), Nvidia (+8.5%, +$2.46B), both Alphabet lines (GOOG +34.7%, +$2.36B; GOOGL +10.7%, +$1.61B), Meta (+21.0%, +$1.86B), Apple (+7.5%, +$1.88B), and Amazon (+5.4%, +$0.73B) are all being leaned into at already very profitable marks.

They are paying up for these names: Nvidia is up 329.9% versus their average cost, Alphabet’s two lines are up 250.6% and 166.5%, and AMD and Micron are similarly deep in the green, yet all see higher share counts. That says they see the AI capex cycle and data‑driven advertising/cloud cash flows as earlier‑cycle than the market fears, not late.

The single most eye‑catching allocation move, though, is EMXC, where shares jump +14053.1% and the position grows by about $2.32B. Combined with sizable adds to SPY (+$2.29B), IWM (+30.4% shares), IJH (+36.9%), and modest increases in EFA and IWD, Goldman is buying broad ex‑US and factor beta instead of sliding further out on individual stock risk.

Behind the headlines, they also quietly scale into second‑derivative AI enablers like Micron (+40.8% shares, +$1.23B) and AMD (+34.3%, +$0.64B), plus semi cap names like KLA (+28.4%) and Applied Materials (+17.4%). That’s a view that AI demand stress will show up as volume and pricing power across the memory and tools ecosystem, not just in Nvidia’s P&L.

The sells are small in number but loud in message: Goldman is taking chips off the table where the AI trade looks fully recognized and freeing liquidity from low‑value wrappers. The standout is Lam Research, where they cut shares -29.4% and pull out roughly -$0.93B, even though the name sits over 300.7% above their average cost.

They also trim QQQ (-7.0% shares, -$0.24B) and IVV (-2.0%, -$0.18B), while still adding to SPY and QQQ’s underlying mega‑caps. That’s a rotation from paying ETF fees for concentrated large‑cap growth exposure toward owning the index via cheaper sleeves (SPY, VOO) and the specific platform names they like most.

Berkshire Hathaway is gently sized down (-2.9%, -$0.13B), signaling less need for a generalized quality defensive now that they’ve built out their own blend of defensives and energy. Morgan Stanley is shaved by -2.2%, a modest expression that bank beta is less attractive on a risk‑adjusted basis than the secular AI and EM stories.

Notice what they are not selling: Nvidia, the software megacaps, or the high‑multiple AI semis like AMD and Micron. The trims are funding sources, not reversals – taking gains in areas where the market has already rewarded them and redeploying into where they see better incremental upside or diversification benefit.

At the sector level, the book is edging, not lurching, but the direction is clear. Technology’s reported weight rises to 48.26% from 47.44%, even after a steep cut to Lam Research, because the capital is being recycled into platform software, cloud, and AI semis with more perceived earnings torque.

Financials drift down to 3.83% from 4.08% as they trim Morgan Stanley and leave JPMorgan and Bank of America as their core bank exposure. Industrials slip to 4.13% from 4.34% despite adding to Tesla and Caterpillar, reflecting how small those bets are relative to the AI complex.

Energy is one of the quiet climbers, up to 3.21% from 3.05%, via meaningful adds to Exxon Mobil (+22.3% shares, +$0.99B) and TotalEnergies (+19.2%, +$0.40B), with Chevron also increased. That looks like a deliberate hedge: if AI and EM growth drive power demand and commodity usage, integrated oils should capture a piece.

Consumer exposure is subtly upgraded in quality rather than quantity. Consumer Discretionary’s overall weight ticks down slightly to 9.44% from 9.60%, but inside that they add to staples‑like names such as Walmart (+13.2%), Costco (+24.1%), Procter & Gamble (+10.9%), and Coca‑Cola (+11.7%, in Consumer Staples at 0.76% weight), balancing the growthier Amazon and Netflix adds.

The “Unclassified” bucket at 21.13% (down marginally from 21.45%) masks a real reshuffle in ETF usage: money moves from QQQ and IVV into SPY, IWM, IJH, EMXC, and VOO. In effect, they’re trading some concentrated US growth beta for broader US, small/mid‑cap, value, and ex‑US exposures.

Viewed as a whole, the quarter says Goldman wants to ride the AI earnings cycle hard while insulating the book with energy, health care, and diversified beta. The persistence of outsized positions in Nvidia, Microsoft, Apple, Alphabet, Meta, Amazon, and Broadcom — all increased despite large unrealized gains — shows they think we’re in the middle innings of AI monetization, not the ninth.

The big adds to Micron, AMD, KLA, Applied Materials, and Palantir argue they expect continued AI capex and data‑analytics demand to ripple through memory, tooling, and software analytics. Combined with scaling into SPY, IWM, IJH, EMXC, and EFA, they are also clearly positioning for a broader earnings upturn beyond the US mega‑cap cohort.

On the risk side, more Exxon, Chevron, and TotalEnergies, plus chunky pharma and managed care (Eli Lilly, Johnson & Johnson, AbbVie, Merck, UnitedHealth all increased) give them ballast against inflation, policy, and rate volatility. These are classic “sleep at night” complements to a very AI‑heavy growth spine.

Going forward, expect more of the same playbook: trim where AI euphoria has run far ahead of conviction (Lam Research is the template), and recycle capital into the platforms and geographies they see as still under‑owned. As long as their 3‑ and 5‑year numbers stay strong, this kind of high‑conviction, theme‑driven rotation is likely to continue dominating their 13F.