What to buy if AI is transformative

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Good morning. Elon Musk announced yesterday that he plans to spend a “lot less” on political campaigns going forward. Tesla investors rejoiced when Musk stepped away from Doge a few weeks ago, sending shares up more than 5 per cent. But they did not react much this time; Tesla shares inched up just 0.5 per cent on Tuesday. The benefits of having an engaged Musk appear to be priced in. Email us: unhedged@ft.com.

The AI equity portfolio

The US — along with most other developed countries — is facing an impending dilemma. Without sufficient immigration inflows, demographic decline is expected. And, though productivity growth is still strong by many measures, many economists and market watchers believe it will stagnate. There are a range of views about what this all means, but it could imply less output, higher inflation and bigger deficits. 

That is, unless AI saves the day. If it can boost the productivity of workers, AI could help countries overcome demographic challenges and keep growth strong. But that is a big “if”. In market terms, it’s the difference between creating trillions in value for investors, or destroying their hard-earned capital. This is not only true of investors in the magnificent 7 AI stocks; as companies such as Nvidia got more expensive, so did the rest of the market, and we suspect AI had something to do with it:

That boost could be from the perceived benefits of AI to the broader economy, rather than the impact of AI on the companies themselves; the P/E ratio of the S&P 500 without the magnificent 7 is now around the same level as before the inflationary surge of 2022. Either way, if AI does not support productivity, that AI halo should fade. Unhedged recently spoke with Joe Davis at Vanguard, who authored a new book about the history of tech disruption and what that implies for portfolios. He put the odds of the bad outcome at 30 per cent to 40 per cent. That would not mean that AI is valueless; though social media sites don’t have a positive impact on productivity, they are great commercial products. But it would mean that AI won’t save the day.

Unhedged is more interested in the optimistic productivity scenario. Specifically, what it could mean for equities beyond the companies currently producing the AI models, creating the inputs for data centres (chips, wires, energy, etc), or implementing AI in other companies (such as Palantir). That scenario may lead to new revenue streams, products and uses we cannot predict yet — which means we cannot invest in them yet, either. What we can trade on is the marginal impacts AI could have on existing companies’ balance sheets. Those benefits could come in two forms:

• Cost reduction: Certain roles will be fully automated, meaning less overhead for the company and higher earnings. 

• Revenue enhancement: Each worker becomes meaningfully or marginally more productive, pushing output and earnings higher.

Most companies would probably benefit from a mix of both. But there are caveats. Cost reduction would result in mass lay-offs and unemployment, with related economic and social issues, if new AI-enabled industries do not pop up. For revenue to be enhanced, there has to be sufficient demand for more output, or prices have to adjust in a way where higher volume leads to higher earnings. Or, quality could go up, and with it prices and demand. 

So what sorts of equities should one own if AI is all its cracked up to be? Most companies in the S&P 500 could see lower costs through automation or augmentation of IT services, call centres, human resources and probably some sales functions, too. Cost reduction should flow through to most bottom lines. 

Revenue enhancement is harder to select for. Manufacturing companies could see some output benefits, but we imagine the biggest impact will be in services. It could be that pricey service companies — financial advisers, banks, consulting firms — are able to reach new, lower-paying customers. Or, those same firms could improve the quality of their services, drawing in new customers at higher prices. Healthcare companies might benefit, too, by reaching new patients at affordable prices or by more ably using their data to provide better care.

It’s impossible to know, and the valuations might be already accounting for these possibilities. Yet, on balance, this supports a value investing approach: bet on the cheap stocks, and pray that they get an AI boost. 

There is one area an investor might try to steer clear of: Big Tech. That may seem strange, as those are the companies that have driven the AI narrative. But as Davis lays out in his book, the first developers of a new technology are not always the most successful, and many don’t even make it to the finish line. Think of Japanese companies in the PC hardware space that lost out when US software companies piggybacked off their products. Or the mass bankruptcies of the early US railroad companies. With a few big players in the market right now, buying the stock of one is essentially betting that it will remain in the AI arena — which it may not. And, even if it does, all the companies that last will be competing against one another, pushing down prices and earnings.

This feels all the more true for the companies that make AI models without an attached platform — for whom AI is the product, not the tool, such as OpenAI and Anthropic. Ever since DeepSeek showed that a model can be made for cheap, implementation appears to be more important than development.

Correction

Yesterday, we incorrectly described the supplementary leverage ratio (SLR). We should have said that the SLR is all of a bank’s tier one capital (common stock, retained earnings, etc) over its total leverage exposure (total assets plus certain off balance sheet exposures). We apologise. Thank you to anyone who pointed this out.

One good read

A conundrum on the pitch.

FT Unhedged podcast

Can’t get enough of Unhedged? Listen to our new podcast, for a 15-minute dive into the latest markets news and financial headlines, twice a week. Catch up on past editions of the newsletter here.