Neuberger Berman Group 13F Portfolio

The biggest adds by dollars tell a clear story: Neuberger Berman is pressing its winners where the earnings power is compounding fastest. Amazon, Apple, Nvidia, and Alphabet (A) together absorbed well over $1.2B of incremental capital, cementing them as the preferred vehicles for AI, cloud, and digital ad growth rather than spreading bets across the tech universe.

• Amazon: A 10.5% share add and a position now at 3.08% of the book says they want both AI infrastructure and consumer/logistics optionality in a single package. At a gain of 132.4% versus their average buy, they’re not “averaging down” — they’re rewarding a thesis that is working.
• Apple: Up 9.2% in shares and now 3.35% of the portfolio, Apple remains their default hardware-plus-services cash compounder, with a 335.5% gain versus cost giving them room to keep pressing without valuation anxiety.
• Nvidia: A further 4.9% share increase in a 4.08% position suggests they see GPU demand as a multi-year, not one-cycle, story — and at a 458.0% gain versus cost this is conviction, not hope.
• Alphabet (A and C): Modest adds of 4.5% and 2.2% in shares highlight a preference for cash-rich AI beneficiaries that also monetize through search and YouTube, not just model infra.

Beyond pure tech, they push hard into liquidity and optionality via a 17.0% share increase in iShares Core S&P 500 (IVV) and a 10.6% add in Vanguard S&P 500 (VOO), together now over $2.6B. A 15.3% increase in Berkshire Hathaway B shares and a 13.3% increase in Johnson & Johnson round out the pattern: use broad beta and quality compounders as the counterweight to concentrated AI bets.

The sleeper high-conviction move is GE Vernova, with shares up 38.9% and value up by about $152.4M. That signals a deliberate bet on grid, power equipment, and energy transition capex as a durable backdrop to the digital story.

If the buys are about deepening the AI spine, the sells are about cleaning up around the edges. The largest trims by dollars are overwhelmingly in semiconductors and adjacent names — not a repudiation of the theme, but a sharp upgrade of quality and duration within tech.

• AMD: The standout funding source. A 41.1% cut in shares and roughly $605.9M in value freed tells you they’re happy to harvest a 225.9% gain versus cost and recycle from a high-beta AI narrative into Nvidia, semicap tools, and analog chips where pricing power and visibility look stronger.
• Micron and Intel: Micron is cut 13.0% in shares (about $156.7M of value) and Intel 11.3% (about $116.8M). Both trims suggest discomfort with memory and PC/server cycles as the cleanest way to own AI hardware, especially when they can own Nvidia, TSM, and the equipment vendors instead.
• Microsoft and Oracle: Microsoft’s 6.8% share reduction, worth about $289.9M, and Oracle’s 9.6% cut look more like risk management than thesis breaks. Both are still large, high-gain positions; selling into strength helps fund the Amazon/Alphabet adds without juicing net tech exposure.
• Mastercard and Home Depot: Mid-single to low-double-digit share cuts here look like classic sources of cash from mature, fully-valued consumer and payments franchises, not a macro call against the consumer.

Around the edges, they also trim Intel-adjacent and mature industrial exposure (Caterpillar, down 6.4% in shares) and small amounts across household names like Costco and TJX. The pattern is consistent: recycle from cyclicals and older software into the parts of the stack they believe will own the next decade.

Sector weights barely budge on the surface — technology slips from 57.97% to 57.06% — but the internal rewire is meaningful. They are methodically swapping out shorter-duration and more cyclical tech for long-duration AI infrastructure, software platforms, and tools.

Within tech, semis are being reshaped rather than shrunk. Big trims in AMD, Micron, Intel, and modest selling in TSM are offset by adds to Nvidia, Analog Devices, Texas Instruments, Applied Materials, ASML, and Lam Research. That’s a clear pivot toward analog, power, and manufacturing equipment as the durable bottlenecks in the AI build-out, and away from commoditizing or more volatile parts of the stack.

Consumer discretionary nudges up from 13.13% to 13.49% as they lean into Amazon, Walmart, and United Rentals, even while trimming Home Depot, Costco, and TJX. The message: favor platforms and capex beneficiaries over pure-storefront or rate-sensitive retail.

Finance stays roughly flat at 5.57% vs 5.63% as they add modestly to JPMorgan and Morgan Stanley but cut Aon. Real “defensive” ballast shows up in small but targeted increases to health care (4.62% from 4.58%, driven by J&J) and in the rising unclassified bucket — S&P 500 ETFs, Berkshire, and GE Vernova together now 7.18% of the book vs 6.40%. Industrials inch up, helped by CSX and Union Pacific adds, tying the portfolio more tightly to U.S. freight and infrastructure spending.

Taken together, this quarter’s moves read like a manager doubling down on a single macro story: that AI and cloud will keep compounding at scale, and the spoils will accrue mostly to a handful of dominant platforms and their critical suppliers. The portfolio is built to own that upside aggressively, but with built-in shock absorbers in quality cyclicals, staples, and broad index exposure.

On the upside, the bet is that Nvidia, Apple, Amazon, Alphabet, and select semicap and analog names keep growing earnings far faster than the market. By nudging up rails, industrials, and GE Vernova, they’re also positioning for a multi-year capex and infrastructure cycle that benefits from AI-driven power demand, reshoring, and logistics bottlenecks.

On the downside, the ballast is clear: larger sleeves in IVV and VOO, a bigger Berkshire stake, more J&J, and steady positions in Eli Lilly, Walmart, and McDonald’s. That mix suggests they want to stay fully invested — not time corrections — but be able to weather volatility in high-multiple tech.

Looking forward, expect more of the same: incremental capital toward dominant platforms and the bottlenecks of AI build-out, funded by trims in lower-conviction semis, mature software, and rate-sensitive consumer names. Unless the AI earnings narrative cracks, Neuberger Berman seems content to run a tech-heavy book, but one that is more about owning toll roads and shovels than every miner in the rush.