This article was produced in partnership with T. Rowe Price
Stockpicker and portfolio manager Peter Bates says the lofty valuations attached to much-hyped AI hardware providers like Nvidia are becoming problematic. But that doesn’t mean the broader AI market is in a bubble.
Bates, who oversees the global select equity strategy at asset manager T. Rowe Price, says he is now underweight companies that could broadly be described as AI “arms dealers” – i.e. they are involved in the manufacture, design and distribution of equipment like chips powering the phenomenon.
“A year ago, Nvidia was one of my biggest overweights, but that was when the stock was 40 or 50 per cent below where it is today,” Bates tells Professional Planner from Baltimore, in the US state of Maryland.
“I think the ‘law of large numbers’ is catching up with it. I still own Nvidia, but Nvidia is an underweight because I think Nvidia is decelerating.”
However, he draws a big distinction between a potential softening in the AI hardware market and the “tech bubble” of the late 1990s coinciding with the rapid rise of internet-based technologies.
The first major difference, Bates says, is that the dotcom boom and bust were largely funded by debt, or at least venture capital equity, whereas capital expenditure on AI in the current market is largely financed by cash flow from a handful of profitable mega-corporations, namely Google, Meta, Microsoft and Amazon. In that sense, he says the AI boom is built upon a much “healthier” capital ecosystem.
The second is that the pace of innovation when it comes to the hardware vastly exceeds that of the broadband equivalents in the dotcom era. “The next generation of chips get better than the last generation,” Bates says. “So, if you’re Microsoft, even if you’re keeping your compute steady, there’s still value in buying the better chips, because they’re faster and more efficient in terms of energy.”
He adds: “To me, this is not a bubble. This is more just a deceleration, and these companies need to grow into valuation.”
‘Penny-wise, pound-foolish’
While his portfolio of about 35 high-conviction stocks is now underweight AI hardware, he is bullish on many of the so-called AI model creators i.e. the household name tech giants who are in pole position to monetise AI, such as Microsoft, Amazon and Meta.
“I do think the rich get richer because there’s such a moat around cost to compute and cost to build these models,” Bates says. “There are probably some private models out there, but the odds are that the big efficiency gains [are] coming from the big guys who are already spending billions.”
He suspects that the boardrooms of big tech giants are preoccupied with the question of what they need to do to win 10 years out, and that, invariably, AI will be central to whatever their answer is to that question. Therefore, any company that takes the decision to cut capital expenditure on AI innovation is “penny-wise, pound-foolish”.
To that end, Bates believes large-cap equities provide a superior mechanism for gaining exposure to AI compared to early-stage venture and private equity.
But while he has conviction that household name tech giants are best placed to take advantage of the AI boom, that doesn’t mean all Magnificent Seven stocks make the cut for his concentrated portfolio.
‘Meta is playing offense, Google defense’
For example, Bates describes Google as a “very good company” that generates “great cashflow”. But he does not own it because it arguably breaches his golden rule: “don’t own companies that can go backwards”.
While he admits the mantra may over-simplify the practice of stock selection, it reflects a commitment to focus on valuation. And Google, he says, faces a very real threat of disruption from challengers such as TikTok in its core vertical of search. “That creates risk in the stock and I don’t think the stock is so cheap that that risk is reflected in the stock price,” he says.
By contrast, he says Meta is well-placed to capitalise on a slew of megatrends including AI and the booming digital ad market, especially in the growing video segment. He says Meta’s AI capabilities allow it to conduct more sophisticated ad targeting than peers allowing it to circumvent headwinds stemming from an increasingly regulated privacy and data sphere.
“I think meta is playing offense with AI, where, where Google is playing defense,” Bates says.
Speaking of defensiveness, Bates warns that traditional diversification across sectors is no longer the hedge it once might have been in an AI-powered world. He offers the example of utilities, which are often seen as very safe, defensive stocks a world away from Silicon Valley growth stocks.
However, some utilities are linked to the AI boom as independent power producers – and even show signs of correlation to market gyrations around AI darlings like Nvidia. “If the utility you own acts like a growth stock, you’re not really diversified,” he says. “I think the last two to three years have kind of exposed the flaw in that mindset.”
‘Picks and shovels, not the miner’
Instead, investors and their advisers should be looking to construct or allocate to a “portfolio of no excuses” which keeps a steady eye on valuations, has a sound risk-reward equation and doesn’t become “inadvertently tilted” towards a particular theme, sector or market.
But some megatrends are worth specifically seeking exposure to within a portfolio. In addition to AI, Bates singles out glucagon-like peptide-1 (GLP-1) drugs, which have been tipped to pave the way for world-changing medical breakthroughs by significantly reducing obesity and related conditions such as diabetes and heart disease.
Though he owns some manufacturers of GLP-1s, and lists Eli Lilly as an example of a stand-out company, Bates also has a firm view that he wants to own companies that are involved in the tools and equipment used in production and development of these drugs. In other words, the “picks and shovels [of the gold rush] not the miner”.
Bates explains: “We don’t know who is going to develop the next great blockbuster biopharmaceutical, but I know who’s going to sell the equipment that’s used when you mass produce it.” He singles out Sartorius, Thermo Fisher Scientific and Danaher Corp as three examples of companies that fit this bill.
“They are not cheap stocks, but they are not expensive stocks for the quality of business,” he says.
