Why we’re buying the AI dip: Asset Allocation Committee views

“There is no insight that AI is cracking.”

“Valuations are not crazy.”

The bulls on our Asset Allocation Committee, including those behind the quotes above, believe the recent weakness in high-momentum AI stocks represents a correction within an intact fundamental story.

Some members also believe we are close to peak oil prices and peak interest rates, which would give support to risk assets. (For now, we continue to hedge these risks, as oil and rate pressures can kill rallies and may trend higher in the near term.)

However, the main debate at our latest committee meeting wasn’t centered on oil or macro conditions but valuation. In the end, we voted in favor of a modest tactical overweight to equities.

In the eye of the beholder

By some measures, valuations look reasonable. The Russell 1000 Growth Index is trading at roughly 21 times earnings,¹ down from a peak near 30.

Other measures look stretched. The gap between the S&P 500 earnings yield and the U.S. 10-year Treasury yield sits near or above dot-com-era peaks, as do the S&P’s price-to-book and price-to-sales ratios. But the S&P’s return on equity is about double its dot-com level. Valuation is in the eye of the beholder!

The most bearish comment from a committee member was that “this level of investment, of capital deployed, is reminiscent of a bubble.” The bears point to “too many parabolic charts.”

The unusual feature of this cycle is the “parabolic” trend in “E.” Earnings have been so strong that the valuation debate now centers on their sustainability.

To infinity and beyond?

During a Bloomberg Surveillance interview late last month, host Jonathan Ferro and I discussed the “infinity and beyond” view of the AI theme (to borrow Buzz Lightyear’s famous line)—namely, that it’s expanding across all parts of the data center supply chain and ultimately to adopters.²

The last four weeks have taken significant air out of AI and momentum stocks. Yet this looks to be a “falling with style” scenario, to stick with the Toy Story reference.

“These corrections are healthy,” expressed one member.

Of course, there are headwinds to the AI trade, such as data center moratoriums. News flow surrounding open-weight models could also drive further volatility, but we don’t believe that open-weight advancements are negative to infrastructure demand. We see the recent pullback as a buying opportunity.

Real demand

The bullish camp argues that there is tremendous demand and enthusiasm for the end product of AI, evidenced by one billion monthly active users on ChatGPT³ and strong enterprise demand for Claude.⁴ They describe AI as a fundamentally “parabolic” technology revolution.

The opportunity set has also been broadening. As models have improved and companies have begun to shift to agentic coding, bottlenecks have migrated beyond leading-edge chips into areas such as networking, memory, advanced packaging, and thermal management solutions.

Ultimately, however, the focus should shift to how companies are leveraging AI. End demand is real, and users are just scratching the surface.

Across our own research platform, we’re making extensive use of AI. Our Integrated Equity team, for example, recently built a large language model (LLM) framework that captured information on the quality factor that was distinct from typical quantitative metrics.⁵ Our analysts are also using LLMs to extract signals from alternative data sources, among other use cases.⁶

I expect that companies across industries will similarly use AI to improve processes.

Infrastructure-related bottlenecks remain a compelling way to play the AI trade. But the users of AI—companies that can harness it to create competitive advantages—should eventually drive the trade.

A diplomatic resolution

At the end of the debate, after the committee had voted to increase its equity exposure, I asked our AI agent, Alex (a nonvoting member), who had been listening attentively, to distill everything she had heard. Her summary of our live discussion was excellent and generated almost immediately—after the humans had spent the better part of an hour reaching a resolution. She was also quite diplomatic, something I find difficult to do as a human!

“From what I’ve heard, you favor adding modestly to stocks. You’re sympathetic to the valuation and concentration worries, but the balance of what you’ve discussed points to resilient growth, easing inflation risk, and a still-constructive backdrop. You support the move, and you prefer to stay mindful of how much active risk you spend on the most crowded parts of the market.”

Final thoughts

A broadening AI theme, continued earnings growth, and the intensity of capex spending underpin our conviction. At the same time, we remain mindful of valuation concerns, market concentration, and risks stemming from the recent escalation in the Middle East.

We are keeping our active risk well below its long-run average and leaving room to adjust as the evidence changes.