Overview
What happens when you gather a technology analyst, a portfolio manager who focuses on value, and another who describes himself as an AI-sceptic? You get a typical day at Fidelity International, captured on the latest episode of Fidelity Answers (listen here).
“I try to spend a lot of time with people who disagree with me,” says Jonny Tseng, our analyst and AI advocate. “I want to know that if I’m wrong, why would I be wrong. But if I can show that their view is wrong, then I know my view is right.”
Tseng is in good company. There are few topics which provoke as much disagreement among investors as artificial intelligence - particularly around whether the immense spend in AI development is justified.
And it is immense. “It’s companies the size of small nation states putting their chips all in on the capex,” he says.
“That tells you that the prize is great. But it also means that the stakes are greater.”
Structural technology, cyclical investment?
We ask our investors how they rate the potential returns from AI over the next ten years. Our value portfolio manager, Rosanna Burcheri, is on the optimistic side: “I give it seven out of ten. We are starting to see more application, the ways in which AI can help in doing things that actually create savings for a company. But we still don’t know what is going to be the killer app.”
That’s a question on the minds of all our contributors. Tseng talks about his recent trip to Silicon Valley, which he visits regularly to meet the companies that are often leading the charge on the next big thing. He explains that last year, there was lots of excitement around the big picture scaling laws for this technology. Now the focus is on how companies convert that scaling into products - “and just in time”.
“We have been waiting some years now to see products which people can put into their hands and generate real revenues.”
In the absence of those products, AI is yet to deliver the sort of returns that investors - and companies - might expect. That’s certainly the view of our contrarian portfolio manager, Dmitry Solomakhin: “The US market has been driven by stories and momentum for a number of years. Parts of the market are in a bubble.”
Solomakhin, like the others, does not doubt the viability of AI as a technology, though: “AI is real and structural. It will continue for many decades.”
His scepticism derives from how investors are responding to it: “The market takes a theme like AI, and takes it out of proportion. It magnifies it, multiplies it, extrapolates it. And that’s how bubbles are made.”
Burcheri agrees that the technology is here to stay, but that in itself does not justify today’s sky-high valuations:
“This is a structural change in terms of technology. But the investment, that’s the bigger question mark right now.”
Seeking escape velocity
How does our AI-sceptic rate the potential returns over the next ten years? “I think there will be a small group who will get a 12 out of ten,” says Solomakhin. “And there will be a big group that get zero. Or maybe negative.”
It is the nature of technological developments like this to create panic among companies, a panic which precedes an inevitable consolidation.
“The motive isn’t just profit for companies - it’s existential,” explains Tseng.
“If your company doesn’t exist anymore, then there is no profit.”
Today’s is a particularly brutal market cycle. Burcheri explains that with previous technologies, there was an advantage in not moving too quickly: “Apple didn’t come up with the first smartphone. But it came up with the best.”
That advantage has disappeared. Tseng describes the problem that companies face with “recursive self-improvement”:
“The idea is that the first companies getting superintelligence will be able to use that to make themselves even better.
“Then you reach escape velocity. Your AI gets a lot better, makes new discoveries, and suddenly no one can stop you.”
Tseng is positive on AI, but realistic about the impact it will have: “Company executives have all read The Investor’s Dilemma, which argues that if you get disrupted, you’re dead. They’ve all seen [what happened to] Nokia and IBM. In the old days, they were kings. Then they got disrupted.”
Harnessing the hype
So where do Solomakhin’s small group of winners lie? Among those who can harness this technology and channel it into tangible applications.
We are starting to see evidence of that. Tseng explains how the companies he met in Silicon Valley are making progress around agentic AI, particularly for customer services, but also incorporating the software into business processes, most commonly to programme code. One company he met uses AI to write 80 per cent of its code.
This represents a “critical unlock”:
“If you can programme code, you can do anything a computer can do.” And you can drive the sort of efficiency gains that proponents of AI have promised from the start.
But this alone is not enough. Hype, momentum, vibes - Silicon Valley has those in abundance. But our AI-sceptic is not a vibes man. He wants numbers.
“If someone improves ad targeting, or customer service, or backend software development, what you want to see is top line inflation or operating margin improvement, or both. When we start seeing these, you can tell this is tangible value, and I can quantify it. I know how to measure that and offset it versus my initial investment. Then, I can justify the return on investment.”
Despite their differences, all our contributors agree that we’re not there yet.
