The biggest risk for US equity investors is not being underexposed to artificial intelligence (AI), but being exposed in the wrong places, according to Baillie Gifford American fund managers Lillian Li and Ben James.
As capital floods into visible beneficiaries of the AI expansion, including data centres, energy and infrastructure, the pair argued much of the real economic impact of AI is already being captured elsewhere in portfolios, often without being labelled as AI at all.
“The market really wants to push the narrative that there is a bubble,” Li said. “But it’s very complex, because different parts of the world are accelerating at different speeds, and from the outside it all looks the same.”
For fund selectors, she said, the more relevant question is whether valuations reflect where durable returns are actually accruing.
See also: AI debt issuance is ‘transforming’ the corporate bond market
The ‘prize commodity’
Li said one reason AI revenues appear elusive is because much of the spend today is directed at foundational work inside enterprises, rather than the end products.
“So much of the current debate is ‘where is the revenue?’” she said. “The answer is kind of very unsexy – it’s all happening in the background.”
She argued most companies cannot extract value from AI until their data is properly organised, something many companies aren’t even close to.
“All of our valuable data, which is a prize commodity of this age, is sitting in completely different sources,” Li said. “The really unsexy thing is getting them all in a format that AI can use.”
This, she said, is already driving spending across cloud platforms and data infrastructure providers, even if it is not explicitly branded as AI investment.
See also: Baillie Gifford: ‘We are appalled by Saba’s actions and conduct’
“When you talk to these folks, everyone’s saying it’s not ‘AI products’ per se,” Li said. “People are spending so much because they’re getting the data in order to use AI.”
By contrast, companies that had strong digital foundations from inception are already seeing benefits feed directly into revenues.
“They’ve had their house in order. They’ve woken up with their running shoes on,” Li said. “They’re telling us we’ve seen improvements across conversion rates… and that’s going to come through as advertising revenue. Not as ‘AI revenue’.”
Over time, she argued, the distinction will fade entirely.
“In the future, we won’t call AI anything,” Li said. “It would just completely diffuse into our lives. It would come through as revenue.”
James added: “Around 95% of our portfolio will have direct exposure to AI from those different layers. But it’s like the internet in 2003 – most companies are starting to use it and gaining traction to really shift their businesses.”
He explained while around a third of the £2.85bn portfolio is exposed to technology, AI influence cuts across consumer discretionary, communication services and other areas.
“It’s not an AI portfolio,” he said. “It’s a portfolio of businesses using AI as a general purpose tool.”
Meanwhile, Li added many of the biggest beneficiaries would never be categorised as AI companies.
“They will still be, for all intents and purposes, a bank or a healthcare company,” she said. “They’re just doing it better with AI.”
See also: Pictet AM’s Lee on AI: Trying to not leave the party too early
Bubble debate
Coming back to the ‘AI bubble’ concerns, Li said it is far too early to be calling it anything like that.
“We are saying that two years in, we are completely over it,” she said. “That feels like very high confidence for something I don’t see in the data yet.”
However, she said selectivity matters, particularly in areas directly tied to capital expenditure.
“Companies related to the capex spend have gone up a lot more than their fundamentals. We’re not exposed to those.”
Instead the fund is focused on “bedrock companies” such as hyperscalers and businesses with strong cash generation.
James said: “If the hyperscalers turned off capex tomorrow, how would that impact their business versus others?
“They have the assets, the models and the distribution.”
Li added: “There is something happening. You just need to know where to look.”
