Commentary: Dividend Strategy

• While the defining characteristic of this market continues to be its profound concentration, cap-weighted and equal-weighted indexes have produced similar returns over long periods of time.
• With investors beginning to ask whether AI represents a bubble, our approach limits the risk from a bubble popping. We value securities based on free cash flow yields, gravitating toward those with asymmetric risk-reward profiles.
• Strategy turnover this year was higher than average as market conditions provided opportunities in companies we have long followed and admired.

2025 marked the third consecutive year of this AI-driven cycle. The market-cap-weighted S&P 500 Index rose 17.9% in the year, while the equal-weighted S&P 500 Index gained just 11.4%. The Magnificent Seven, this market’s nucleus, rose 24.9% in 2025.

Since ChatGPT launched three years ago, the cap-weighted S&P 500 has delivered nearly twice the gains of its equal weighted peer (Exhibit 1). The Magnificent Seven, meanwhile, has surged a staggering 332%.

Exhibit 1: Pre- and Post-ChatGPT Returns

The defining characteristic of this market continues to be its profound concentration. Indeed, it is the most concentrated equity market in American history (Exhibit 2).

Exhibit 2: Weight of Top 10 Stocks in S&P 500

Against this backdrop, it is worth noting that while performance comparisons will vary based on the time period, cap-weighted and equal-weighted indexes have produced similar returns over long periods of time (Exhibit 3).

Exhibit 3: Equal-Weighted and Cap-Weighted Index Performance

In 2025, and throughout the last three years, the ClearBridge Dividend Strategy has enjoyed healthy gains, in part due to its investments in AI participants like Alphabet, Broadcom, Meta Platforms, Microsoft and Oracle. We have not gained as much as the cap-weighted S&P 500, our benchmark, however, as we maintain diversification to prevent the portfolio from becoming overly concentrated. Relative to the equal-weighted S&P 500, which is regularly rebalanced to eliminate concentration, but which is not our benchmark, our significant holdings in technology leaders, among other things, drove superior performance in 2025, and over the three-, five- and 10-year periods.

Since launching the Strategy in 2003, we have generally limited individual holdings to 3%–5% of the portfolio. By contrast, today the four biggest companies in the cap-weighted S&P 500 each represent 5%–8% of the index.¹ We typically cap sector exposures at 15%–20% of the portfolio. The information technology (IT) sector currently represents 34% of the market. Throw in Alphabet, Amazon and Meta (leading technology companies not included in Standard & Poor’s IT sector) and that weight rises to a staggering 46%.

The S&P 500 managed a 2.7% gain in the fourth quarter, while the Strategy captured roughly half of that. On the plus side, consumer staples holdings such as Nestle, Coca-Cola and Unilever and a health care overweight shined as concerns of a tech bubble grew.  On the downside, we are underweight Alphabet and Apple, two Magnificent Seven stocks that performed well, and overweight Oracle, which gave back some of its large gains from earlier in 2025 (more on this below). Increasing competitive dynamics in the wireless industry also weighed on T-Mobile.

AI will radically change our lives, the labor markets and the economy. That does not necessarily mean, however, that all (or even most) AI stocks currently represent good investments. Investors already ascribe trillions of dollars of value to AI-related enterprises, yet aggregate AI-related revenues are de minimus relative to the expectations embedded in these companies’ valuations. AI-related revenues are growing quickly, but the landscape is evolving too swiftly and too dynamically to conclude that today’s favored players will be the ultimate winners.

Significant fundamental questions remain, with enormous ramifications for the companies involved. Will large language models (LLMs), like ChatGPT and Gemini, become commoditized? Will America’s expensive approach of developing proprietary LLMs be upended by China’s far cheaper tactic of building open-source models? As revenues from training — which today predominate — plateau, will revenues from inference be enough to keep the momentum going?² Will Nvidia’s GPUs retain their preeminence, or will alternative semiconductors make inroads?

With disruptive technologies like AI, it is critical to distinguish between technological and social impacts, on the one hand, and financial outcomes, on the other. AI will almost certainly be a game changer, but that does not guarantee that Nvidia’s or ChatGPT’s valuations, for example, will ultimately be justified by their cash flows.

With folks beginning to ask whether AI represents a bubble, our approach limits the risk from a bubble popping. We value securities based on free cash flow yields, gravitating toward those with asymmetric risk-reward profiles. We size the investment relative to the range of outcomes, ensuring no individual position or sector becomes too outsize.

Our analysis requires a forecast of upside and downside scenarios and a probability assessment of those outcomes. It is inherently easier to make predictions in sectors that are stable and subject to little change (think businesses with recurring revenues and little risk of disintermediation). It is much harder to predict outcomes in areas undergoing significant, fast-paced change (like AI). Unsurprisingly, we favor companies operating in stable markets and that we can underwrite with a higher degree of conviction.

With several leading AI companies now sporting multitrillion-dollar valuations, it is clear they reflect lofty expectations. As their prices rise and valuation multiples expand, the investment case evolves and can become less favorable. A brief review of our investments in Oracle and Broadcom illustrates how we approach this type of environment. While both stocks surged in 2025, we significantly reduced our position in Oracle, while largely maintaining our position in Broadcom.

Throughout the first nine months of the year, the market cheered each time Oracle announced a new, large data center contract. By year end, however, investors worried whether Oracle’s huge construction backlog represented too much of a good thing. These data center contracts require the company to spend hundreds of billions of dollars, a potentially risky endeavor that represents a profound shift in its business model.

Oracle grew up as a software company, enjoying high profit margins and returns on invested capital, due to the negligible capital requirements of the software business. The AI data center business, by contrast, is phenomenally capital intensive, yet offers lower margins and returns. Further, despite Oracle’s size, the required expenditures strain its balance sheet and raise concerns about the sustainability of its investment-grade credit rating. With its shares surging despite this increasingly complex outlook, we meaningfully reduced the position to reflect the evolving risk-reward. We continue to hold a modest position, however, as we balance the risks of the company’s business model evolution with the potential for years of extraordinary top-line growth.

In contrast, while Broadcom also soared in 2025, we did not radically reduce our exposures. The company’s leadership as an ASICS chip provider positions it as one of a few potential competitors to Nvidia. While Nvidia continues to dominate the AI compute market, customers are flocking to Broadcom to diversify their supplier base. Because these ASICS chips are Broadcom’s core competency, these sales yield increasing profit margins and returns for the company. Broadcom shares could still disappoint investors if competition increases, pricing decreases or volumes underwhelm, but its overall risk is likely still lower than Oracle’s. Broadcom’s AI strategy plays to its core strengths and does not require substantial, incremental capital outlays.

Given AI’s centrality in today’s markets, it feels appropriate to focus our comments here on it. But given our approach to investing — conservative, diversified and risk-averse — AI is not central to our portfolio. Indeed, no one sector or trend will ever be central to the ClearBridge Dividend Strategy. That is the whole point of diversification. It may sound trite, but we will never put all our eggs in one basket. In momentum-driven, concentrated markets, diversification inherently impacts relative performance comparisons. Of course, it also limits losses when that trade, inevitably, unwinds. That we feel compelled to extol the virtues of diversification reflects just how distorted market psychology has become (Exhibit 4).

Exhibit 4: ClearBridge Dividend Strategy’s Diversification by Sector

With the market obsessively focused on the next AI data point, we see exciting prospects in other corners of the investment universe. Indeed, our turnover this year (29%) was higher than average (15% over the last five years). Market conditions provided opportunities in companies we have long followed and coveted such as: Automatic Data Processing (a fourth-quarter addition), Inditex, L3Harris, Marsh & McLennan, Old Dominion Freight Lines and TE Connectivity. We also took advantage of situations in select existing holdings to significantly increase our positions in Air Products, Exxon Mobil and Union Pacific. All represent high-quality franchises in attractive industries underwritten based on reasonable cash returns. It is an eclectic bunch, sourced from bottoms-up analysis, not a trend-following, top-down approach.

Looking ahead, we anticipate the AI debate will carry over in 2026. We will continue to participate in AI in a measured and disciplined way. Yet, as other investors continue to myopically focus on AI, we expect to find additional idiosyncratic opportunities in overlooked corners of the market. We believe these seeds will bear fruit in the years to come.

The ClearBridge Dividend Strategy continues to trade at a significant discount to the broader market. Our average holding has grown its dividend at 10% over the last 12 months; we expect similar growth in the years ahead. With market concentration and valuations at or near all-time highs, we remain committed to diversification and valuation discipline. These principles have served investors well in the past; we believe they will continue to serve us well in the years ahead. 

Exhibit 5: ClearBridge Dividend Strategy Trading at a Meaningful Discount

The ClearBridge Dividend Strategy underperformed its S&P 500 Index benchmark during the fourth quarter. On an absolute basis, the Strategy saw positive contributions from seven of 11 sectors: the health care sector was the main contributor, while the real estate and IT sectors were the main detractors.

On a relative basis, underperformance was driven primarily by stock selection in the communication services, IT, materials and real estate sectors, while stock selection in consumer staples, a health care overweight and IT underweight proved beneficial.

On an individual stock basis, the main positive contributors to relative returns were AstraZeneca, Inditex, Freeport-McMoRan, Merck and not owning Netflix. An underweight to Alphabet, Ely Lilly (not owned), T-Mobile, Enbridge and Micron Technology (not owned) were the main detractors.

In addition to the purchase of Automatic Data Processing, during the quarter we initiated new positions in PG&E in the utilities sector and Roche in the health care sector. We eliminated positions in Edison International in utilities and Merck in health care.