Why a high-cost world should favour stock pickers

Robert Almeida is looking past AI or at least looking at what’s underneath it. The Portfolio Manager and Global Investment Strategist at MFS Investment Management recently penned an outlook which posits a broader macro shift in investment regime that AI is more a component of than a driving force within. Since 2022, he says, the investment regime of low inflation, low interest rates, cost cutting, and strong index performance that defined the 2010s is now over. What has replaced it is a high cost, high inflation, high rate environment where businesses will win and lose on their capacity to spend capital, earn real cash flow, and price making capacity.

Almeida explained that while AI is part of that narrative, it comes with its own elements of narrative “ephemera” that could distract investors from that more essential core dynamic in markets. In the short-term, AI adds to the inflation story and pushes costs higher, even if its productivity promises imply long-term deflation. He notes, too, that the pre-COVID investment regime prioritized corporate efficiency in the Untied States so much that, even with AI, many companies are finding it harder to cut even deeper, forcing them to focus on how they can establish themselves as price makers. 

“What I am trying to challenge is that every industry is going to be facing new vectors of competition that it didn’t face before. I think we’re running to a world where it’ll be less about earnings growth, will be less about how much cost you were able to rip out of the business through cheap labor in China. It’s going to be more: do you have a product that’s mission critical, that can’t be easily substituted. Companies will pay for that and you’ll be able to offset these price pressures,” Almeida says. “In that world, we shift to more of a stock pickers environment across a variety of sectors.”

While in the short-term, Almeida notes that pricing power currently lies in AI hardware components like GPUs and memory, he says that will eventually end when the supply of AI compute eventually exceeds demand. We don’t know when that will be, he notes, but when it comes that period of pricing power should end.

While much has been made about AI’s capacity to replace human employees, Almeida casts some doubt on the idea. He argues that AI is a stochastic parrot, with amazing capacity to analyze existing information but zero emotional intelligence, human judgement, or the capacity to say whether the question it’s being asked is the right question in the first place. That is not to say Almeida believes certain tasks or jobs won’t be automated by AI, but he believes that the businesses that can hold on to human judgement will succeed.

“The risk is for businesses that go all AI will be guaranteed to produce something that’s mediocre. And if it’s mediocre, then returns are ultimately going to collapse to the marginal cost of capital because it’ll be a commodity,” Almeida says.

There is a risk, Almeida notes, that companies who see AI as a replacement for most of their workforce will suffer from the ‘doorman fallacy,’ an idea espoused by British advertising executive Rory Sutherland. Sutherland notes the example of doormen in New York City apartment buildings. Their primary role, opening a door, can be done by a very simple machine: a sliding door or a revolving door. When buildings fired their doormen in favour of an automatic door, however, they lost the person who greeted residents, made sure packages went to the right unit, kept unsavoury characters away from the building, and elevated the prestige of the building they served.

To Almeida, there is a risk that companies are so motivated by their need to report quarterly growth, investment in AI, and cost cutting measures to serve the needs of investors that they end up cutting too deep. Some of the efficiencies these businesses hope AI can bring them may not manifest, leaving them in a situation where they have to replete their stocks of skilled labour.

Once AI’s promises are reconciled with its realities, which will be explored in the second part of this series, Almeida believes that the new test for a stock will be on the quality of its cash flow. If a company can sustain pricing power and pass through rising costs, then those businesses should be strong assets to hold.

Almeida sees more promise outside of the United States, in part due to that core cash flow metric and in part because of lower valuations that reflect and eventual mean reversion in profit margins that he believes is coming for US stocks. In addition, he sees greater capacity in markets with less efficient businesses to add value through AI. Likening AI to GLP-1 drugs, he says that the companies that can best use AI are the “labour obese, mature, sleepy organizations outside the Untied States.” The US focus on efficiency means that its businesses have already torn out so many costs, making it a market less primed for AI cost savings.

Almeida sees a host of challenges ahead, and potentially violent corrections that could come as AI promises are reconciled against their realities. He believes, though, that advisors can prove their worth in the moments that lie before them.

“My message to advisors is really just a subconscious or subliminal commercial for discretion and advice, which is the stocks that have gone up the least are the ones that hopefully will have more economic viability versus the ones that have gone up the most are the end of the bullwhip in the in the AI trend,” Almeida says. “Eventually you’re going to have enough memory. You’re going to have more memory than you need. Eventually you’re going to have more chips than you need. What matters whether the company is providing a tangible value add to its customer, and whether its customer is willing to pay for that. I don’t know what the end market looks like, but I think you get really rewarded for owning the winners and you get put out of business if you own too many Lehmans.”

A second piece, outlining Almeida’s view on a possible correction in the AI trend, will follow