Huge investment in AI by the Magnificent 7: Fibre optic cables again?
Companies such as Global Crossing and Nortel had grand plans and huge investments in fibre optics. Yet they went bust in the noughties.
Is the hype around AI creating a similar overinvestment bubble? “In some areas such as language models, [the answer is] yes,” affirmed Mozamil Afzal, global CIO and CEO at EFG Asset Management, during an interview on 26 February 2025. “If it is indeed open source, there’s no value there, because everyone has access to it.”
“You’re always master of none? You never do one thing well”
Afzal is also sceptical of companies involved in the full artificial intelligence value chain (hardware, cloud infra, models, apps, customers) such as Google. The company may have become the 2025 version of IBM in the 1970-1980s. “Back in the day, IBM was doing everything for everybody, and then it disappeared on everything,” said Afzal.
I want to see companies that are either tangibly cutting their costs using AI or have improved the customer experience
He wonders whether Google should not be broken up to create value for shareholders. These very large IT companies have become “average on everything” relative to competition. “Amazon’s AWS is fantastic in AI infrastructure and big data, they’re the best in class for that one thing.”
Could Google have created a powerful ecosystem similar to Apple?
“It could that could well be the case. I’m willing to bet against that… because history doesn’t tell us that,” firmly affirmed Afzal. He thinks that Apple may not be a good example as he claimed that the innovative company makes 85%-90% of all its profit from phones.
AI: Business models that may work and support stock valuation
“I want to see companies that are either tangibly cutting their costs using AI or have improved the customer experience,” analysed Aftal. “Meta, in my mind, is in the lead.” He thinks that Meta’s positive equity performance this year reflects the ability of its customers, which are B2B, to get a better ROI, using social media firm’s AI. Besides, Meta recently fired 5% of its staff as AI “is just as good and probably codes faster.”
Consequently, he believes that the markets will likely be more discerning in the second half of the year by asking whether AI has added value to clients. “Everyone puts AI next to their company name, like in 1999 [during the dot com bubble] and then the share price just goes up. If you’re not [adding value], those stocks will tumble very quickly [as in 2000-2001].”
Is AI adding more value in the creative industry?
“In the entertainment industry, you can get to market with a movie much faster. That’s going to reduce the cost of the whole industry.” He thinks that AI will enable exponential creativity. Yet he also thinks that what you do with your hands as artisanal [will gain] much more value… than what will be produced by AI.” Ferrari is a good example where creativity is at its best and it is reflected in the highly valued share price.
“To me, [creativity] is a really good example of how Europe actually can differentiate.” During the interview, Afzal struggled to quickly identify creative US companies apart from Tapestry with its Coach handbags. “Even Tiffany got bought by LVMH.” He thinks that Europe is home of “real artisanal creativity.”
Nvidia and Google were not built in a day
Despite being an industry defined by its disruptive nature, as observed with Google with its innovative search algorithm in the early 2000s, Afzal would not invest in European AI firms in the early part of the AI value chain as “they have missed the boat… the game is over. We're going to use Nvidia chips,” argued Afzal. He thinks that the monopoly of ASML in lithography is rather more the exception than the rule.
A newly created European chip company could “never win.”