Energy Hedges the AI Trade

An Index Behaving Like a Single Stock

A lot of life, and certainly the challenge of portfolio construction, is about tradeoffs. Portfolio construction has a key constraint: many of the tradeoffs emerge because of uncertainty, with one of the main ones being that we are looking for signals in a very noisy environment. Even when we get the signal right, it doesn’t always translate as expected at the portfolio level or takes longer to cut through the noise than we thought. This matching problem between signals and outcomes directly lowers your edge at a given time scale and is why diversification is such a critical tool in portfolio construction: concentration is only justified when edge is high, but when edge is reduced the solution is to diversify more.

Right now, indexes are increasingly concentrated, effectively behaving as if they have significant edge, and record-low correlation, coupled with the “stock up, volatility up” dynamic, has them acting more like a single stock with major upside skew. Managing relative risk against this dynamic is extremely challenging: how do you manage against a single stock with a diversified portfolio? Especially when we think the actual edge is low, correlations cannot sustain current levels, and equity risk premiums are at levels that have historically called for caution.

The result is an incredibly steep tradeoff between managing absolute and relative risk—arguably the steepest since the 2000 cycle. Our investment process is anchored by managing absolute risk through an absolute valuation discipline, with our primary goal being to generate a double-digit absolute return across a full market cycle. We do this by exploiting price-to-value gaps specifically where we expect convergence because future fundamentals have a higher probability of a good outcome than the market has priced. We then look to manage the resulting volatility tax with differentiated portfolio construction, combining stocks across a much wider correlation spectrum than most peers would consider.

This entire regime depends on correlations staying at extremes. The only other times realized correlations were this low were just before the 2007 housing crisis and just before “Volmageddon”¹ in early 2018 (Exhibit 1). The likely driver of this low-correlation extreme is AI: there is AI and everything else. The paradox is that while the ultimate outcomes for AI are inherently uncertain, the market is behaving as if it has a lot of edge around these AI unknowns. Ultimately, we believe that correlations will spike higher, and our guess is the timing will be measured in quarters rather than years.

Exhibit 1: Indexes Are Seeing Record Low Correlations...Again

As of September 7, 2026. Six-month realized correlations for the S&P 500 Index. Source: Bloomberg. Past performance is not necessarily indicative nor a guarantee of future performance.

Another key driver of this low-correlation market is record dispersion in stock betas, emerging not only across sectors but within them (Exhibit 2). If you don’t manage this dispersion directly, the gap in beta between a prudently diversified portfolio and the index creates an active beta drag—a relative risk tax.

Exhibit 2: Sector Betas Are Seeing Record Dispersion

As of August 31, 2026. Source: Bloomberg, ClearBridge Investments. *Note: p10 / p90 of Member Betas = 10th / 90th percentile of the individual member betas within each sector of the Russell 1000 Value Index.

The Ultimate Risk Hedge in an Energy Crisis

What we have not done is sacrifice the diversification that manages our absolute risk. Rather, we believe this record dispersion provides a great opportunity for active managers, and for us it starts with energy. With the beta of tech at roughly 1.2x and energy at 0.6x, energy’s risk behavior is even more extreme if we look at the past year, when the sector’s correlation to the index has been below –50% and its correlation to the portfolio roughly –20%.

This highlights energy as the ultimate risk hedge in, well, an energy crisis. When energy spikes, pretty much everything else goes down. What amazes us is that we are still in an energy crisis—one that gets worse as the cumulative probability of a bad outcome rises with each passing month—yet the market has priced this crisis as largely solved, despite the feedback from higher energy prices into interest rates. Historically, the only areas of the market that do well when energy and rates rise together are financials, up to a point, and real assets, more consistently. Yet investors are broadly making very little effort to manage this risk as the AI trade dominates all available attention. The key is that owning energy covers our absolute risk while creating the risk budget to cover the AI relative risk more effectively.

As the great Charlie Munger always emphasized, “show me the incentive and I will show you the outcome.”² Despite higher energy prices, capital spending on energy, and on commodities broadly, has stayed extremely disciplined (Exhibit 3). After the shale boom and bust, energy executive compensation shifted from growth-based—which is insane in a commodity business—to returns-based. Ironically, major tech companies’ incentives are focused almost entirely on growth, not returns.

Exhibit 3: Energy Capital Discipline Holding Despite Higher Prices

As of December 31, 2025. Source: The Crude Chronicles. Note: Consensus estimates are used for 2026-2030 projections. There is no assurance that any estimate, forecast or projection will be realized. Past performance is not necessarily indicative nor a guarantee of future performance.

At least for now, the market is treating the AI investment boom as if it is completely insulated from a global energy crisis and higher interest rates, despite the increasing use of debt financing to keep the show going. What is surprising is that the AI trade also seems to be ignoring mounting political risk around data center expansion and increasingly loud warnings from the frontier labs themselves about existential risk from AI. Anthropic CEO Dario Amodei has repeatedly argued that the industry must “pace the frontier,” slowing capability gains before recursive self-improvement outruns our ability to control them. We are not sure what risk premium the market should assign to such tail risks, but investors may want to start considering one.

Unstable Extremes Favor Energy Exposure

Energy stocks have never been this negatively correlated to the market, but we believe these correlations will inevitably turn positive again. The key, as always, is context: we think energy stocks are much more defensive than in the past, when terrible capital discipline destroyed massive amounts of capital through excess investment into downturns, exacerbated by much weaker balance sheets. Today’s reality is the complete opposite. The key risk scenario, however, is that energy prices spike in the coming quarters to levels that cause a global recession. On the way to that level, energy stocks would arguably be the only truly defensive sector, but we would likely reduce our energy weight and pivot to a more defensive position as recession risk increased. Even then, energy stocks still embed oil prices anchored in the US$70s, which we think is a downside floor, as global oil inventories and geopolitical buffers will have to be rebuilt when the crisis finally ends.

Thus, we think the ongoing bull market in real assets that emerged from the COVID cycle has, if anything, been strengthened by the crisis. We expect the energy weight in indexes, which remain near historic lows, to keep rising over the coming years as energy proves to be an investment while many parts of AI prove to be a trade.

The tradeoff for active managers between relative and absolute risk has never been steeper, as diversifying to manage absolute risk increasingly comes with higher relative risk against ever more concentrated indexes. Ultimately, many of the current extremes, especially correlation, are not sustainable, and we think the inevitable journey back to normal will greatly favor disciplined active managers—with energy and real assets covering the flank the AI trade leaves exposed.