The S&P 500's seven-quarter streak of double-digit earnings growth has most investors feeling covered. Michael Landsberg, chief investment officer at Landsberg Bennett Private Wealth Management, says the real risk is already inside the portfolio: AI-adjacent exposure dressed up as sector diversification, and most investors have no idea how deep it runs.

Landsberg's argument is straightforward and uncomfortable. Utilities, industrials, REITs, even some energy names now trade as data-center or AI plays. "That's not true diversification," he said this week. He estimates real technology exposure in the S&P 500, once communication services and Amazon's placement in consumer discretionary are folded in, sits closer to 45 to 50%. For a retired or near-retirement client, that dosage is too high.

Where the earnings are going

The case for rebalancing, Landsberg argues, starts with dosage. Nvidia remains Landsberg Bennett's largest single holding. The firm first bought shares in January 2019 for the GPU and graphics business, before parallel processing and AI reshaped the thesis. Today Nvidia sits at 2.5% of the portfolio, against the roughly 8% weight it carries in the S&P 500. AMD, added in the last couple of years, and Broadcom, held since January 2015, round out the chip names.

The rotation out of concentration points to memory and international. Landsberg flagged Micron and SK Hynix as names where supply cannot keep pace with demand. Micron has pulled back roughly 30% from its highs, putting it near his 10% dip threshold as an entry signal. Apple, he noted, has discussed sourcing memory chips from China because supply from Micron, SK Hynix, and Samsung is not there. On the international side, Landsberg Bennett carries roughly 25% international exposure for most clients, against a typical wire-house allocation of 6 to 8%.

The earnings screen behind that preference: of the top 200 fastest-growing EPS names on the FTSE All Country World Index over the next five years, a large majority are international or emerging market companies. Only one Mag Seven name appeared on that list, which Landsberg identified as Nvidia. Poland, Belgium, New Zealand, and Israel's Tower Semiconductor are among the names he mentioned.

The counterargument

The counterargument is the one Landsberg himself named: most visible international indices carry their own AI concentration. Taiwan Semiconductor and ASML, which Landsberg described as among the world's largest index holdings, trade in close correlation with Nvidia and the broader Mag Seven. An investor buying a standard international ETF may not get the diversification they expect. The line to watch is whether emerging-market and smaller developed-market names, Poland and New Zealand rather than Taiwan, can sustain the EPS growth that justifies the rotation.

On balance, the setup Landsberg described is an earnings story. He projects an 8 to 10% return for the S&P 500 in the back half of 2026, putting the full-year figure near 20%. The September Fed meeting is the friction point: he puts the odds of a 25-basis-point hike at roughly 50/50 if the August CPI print, due before that meeting, comes in 25 or more basis points above current levels. Tesla and SpaceX didn't make his buy list. On SpaceX specifically, his word was "overpriced."

Related reading