Agentic artificial intelligence (AI) cannot both disrupt all software companies and simultaneously boost the companies benefitting from the capital expenditure required to build it out.
This is the view of Kent Hargis, chief investment officer of Strategic Core Equities at AllianceBernstein, who suggested the recent broad sell-off in software stocks may be overlooking a key point.
“In the long run, for the capex to work you have to get paid for it,” he told FSA in an interview. “You need to monetise it.”
“There are many winners in the monetisation of AI and all we seem to be talking about today is that they [software companies] will all be disrupted.”
“There will be winners and losers on the capex side and on the monetisation side,” he said. “Both sides are important, it’s not all winners on one side versus the other.”
Software stocks have endured a major drawdown on fears that agentic AI will erode their businesses, while semiconductor stocks have continued to rally on surging demand for the chips powering this new AI technology.
But Hargis (pictured) points out that even though agentic AI will increase the flow of data and the amount of transactions going through AI infrastructure, the disruption to software companies will not be as quick and easy as some suggest.
“Disruption is not universal and immediate,” he said. “During all previous disruptions, there are winners and losers, and it takes a while.”
“We don’t have the power to supply all of this capex. There are organizational and human behavioural frictions that keep this from happening immediately.”
“On the software side, not every company is the same, just like in the past with Amazon or previous disruptions.”
“We’ve been here before”
Hargis recalled how some investors believed retail businesses would be completely disrupted by Amazon, which was partially true, but not to the extent that some expected.
In the end, a large number of companies ended up being quite resilient and in some cases even more successful.
“Walmart now trades at a higher valuation than Amazon,” Hargis said. “We can give other examples in our portfolio: McKesson for drug distribution, Autozone for auto parts retailing.”
He also pointed to how many companies have tried and failed to disrupt payment incumbents Visa and MasterCard.
“Everybody who tried to disrupt Visa and MasterCard eventually realized it was better to work with them, and they have actually grown their value-added services and their alternative payment rails,” he said.
That is not to say Hargis is sitting on his hands and holding on to companies that the market has sold off.
He told FSA that he has recently reduced exposure to software names where there is some “terminal value risk”, as well as some infrastructure names in the $7.4bn AB – Low Volatility Equity Portfolio strategy he manages.
One recent sale has been in Oracle, which Hargis argued “had gone from a quality company to a company that was becoming highly leveraged with weak cash flows”.
Although quality stocks have faced a recent style headwind, underperforming for the first time in over 20 years, Hargis does not appear too concerned: “We’ve been here before,” he said.
From the frying pan into the fire
When it comes to portfolio construction, investors looking for ways to diversify their US or AI exposure should not simply rotate into emerging markets given how correlated they might be to the AI trade.
This is according to David Wong, senior investment strategist and global co-head of equity business development at AllianceBernstein.
“When I hear that some clients in this part of the world are addressing their over exposure to the US and AI by buying emerging markets, I question whether they’re just jumping from the frying pan into the fire, because there’s plenty of AI exposure in Asia,” he said.
Indeed, the largest five constituents which make up roughly 30% of the MSCI Emerging Markets Index are all AI-related names: TSMC, Samsung Electronics, Tencent, Alibaba and SK Hynix.
Wong recalled during his time as an analyst in the late 90s covering TSMC, Samsung and SK Hynix that he was told by these firms that “demand for these chips was virtually unlimited, that there was no end in sight to the capex plans that these companies had, because the demand was going to be secular in nature”.
He said: “But we also know that semiconductor fabrication facilities are inherently cyclical and you cannot assume that these companies have become suddenly less cyclical.”
The AB – Low Volatility Equity Portfolio still holds semiconductor stocks in its top 10, namely Nvidia, Broadcom and TSMC, albeit at a relative underweight compared to the broader MSCI World index.
As investors gravitate toward asset heavy, low obsolescence businesses in the wake of AI disruption, Wong warned: “They will want that…until they see the next cycle turn down.”
“That is a helpful anchor for us,” he said. “You don’t look at the quality of these companies at their peak level, you have to look at it on a normalised basis, and that normalisation is not something that the market is thinking about today.”
“We also know we need to own some of these companies — the best ones — but we need to manage our exposure to them.”