What comes after the chips trade for AI investors?

It may be unhelpful to see the AI trend as a single trade, the way we look back on dot com stocks from the ‘90s. As artificial intelligence has rippled across society and gripped global equity markets, its role in alpha generation has already shifted multiple times. At first, investors focused on the company’s building AI infrastructure, the so-called ‘hyperscalers’ many of whom were constituent parts of the ‘magnificent seven’ group of companies. More recently, however, that focus has shifted from the companies building AI to the companies those hyperscalers are buying from. Chips, be they GPUs, CPUs, or DRAM became the watchword as investors looked for the picks and shovels of the AI gold rush. A view of the hardware market as cyclical, however, has investors wondering where they should look to next for AI alpha.

Brendan Livingstone, Senior Vice President for Total Portfolio Management at Connor Clark & Lunn Private Capital in Toronto, believes that the question of when, and how, the chip trade ends is one of our contemporary million dollar questions. He notes that the massive jump in earnings that chip making companies have experienced may create unrealistic growth expectations on the part of investors going forward. If the AI buildout reaches its endpoint, or if financial conditions change its pace, then the chip trade might unravel and the next area of AI value has to be identified.

“You can be on a secular growth trajectory but still have cyclical behaviour along the way. And I still think that we are very much in a period where from a cyclical standpoint, the economy is, generally speaking, in decent shape. But economic slowdowns are always going to be a feature, and when that happens, you’ll see a pullback in some of this discretionary spend,” Livingstone says. “There is a risk here that from a cyclical standpoint, we could see a pullback.”

Reassessing quality to find AI opportunity

Livingstone argues that AI has already upended our idea of what constitutes a quality company. Before the rise of AI software was a sector that dominated quality metrics. Software companies had, and in many cases still have, strong profitability, high margins, and tended to have strong protective moats around their business models. AI’s capacity to code and create new software has, in the minds of investors, filled in that moat causing the software sector to correct significantly downwards. AI has the potential to radically change the competitive landscape for companies, especially those seen as higher quality.

Now Livingstone believes that AI’s value can be best realized by companies with two defining traits: the capacity to add AI to improve margin and competitive moats that AI cannot bridge. Those companies tend to be value names, and in some cases were viewed as a value trap for decades.

Livingstone’s prime example of a sector with these traits is the Canadian banks. The banks, he says, have huge amounts of data and can use AI to streamline the labour-intensive processes required to digest, analyze, and use that data. That should result in cost savings and bottom-line improvements. Moreover, the Canadian banks’ competitive moats are based on scale, brand presence, and regulation rather than technology, making them better insulated from AI disruption. He notes that some of these characteristics have already been price in to Canadian banks, with the S&P/TSX Composite Index Banks (industry group) up over 60 per cent in the past 12 months and up nearly 30 per cent in 2026 so far.

Finding, and explaining, long-term AI ROI

While sectors like Canadian banks may be seen as new beneficiaries of the AI trend, there are ongoing questions about how much AI services will eventually cost and how ROI in AI services will be measured over the longer-term. Livingstone says that markets have largely shifted from the view that the promises of future profitability from AI justifies investment now. That view initially buoyed the hyperscalers, but has been replaced by the consensus that companies must demonstrate how AI is improving margin or driving growth for them now. Livingstone, also notes that the rise of cheaper AI models out of China as well as efficiency improvements from US AI service providers may keep costs down and allow for AI profitability improvements to play out over the longer-term.

For advisors, there can be an element of narrative whiplash that comes with the AI theme. The shift from hyperscalers to chip manufacturers to non-tech value names may be a lot to take on, let alone explain to clients in the context of a 30-minute meeting. These rotations have been violent, too, introducing volatility into portfolios that need to be managed. Livingstone says that while he wants to manage opportunities related to AI, he needs to manage risks as well.

“One has to be a bit careful that they’re not taking such a large implicit bet from an AI perspective because the momentum is so strong as well. You have to keep a very close eye on that from a risk management perspective,” says Livingstone. “We want to capitalize on it for our clients because we believe it’s a secular structural theme. But, there are parts of the market now that look like they are a bit frothy. So we have to have valuation discipline, and we also have to have risk management discipline. The spread between winners and losers right now is massive. So you have to be very careful from a risk management perspective that you’re staying within your risk limits.”