One of the hottest picks on the stock market lately is AI. (And no, ChatGPT didn’t write this — a human did.)

Between 2015 and 2022, investment in AI increased seven-fold, and some analysts have even lumped the technology into the “essentials” category, alongside the internet or electricity. With many Fortune 500 companies making big investments in AI, including Alphabet and Microsoft, it would seem everyone wants to play the AI game right now.

…or do they?

Although we’re in the midst of an AI investment frenzy, some insider voices have hinted that the ultra-positive sentiment around it may be starting to wane. Some are even comparing the AI boom to the ‘90s dot-com bubble. (Which, if you didn’t know, left millions of investors holding worthless shares virtually overnight.)

Many AI companies may be over-promising and under-delivering, says Mitchel Green, a well-known venture capitalist. And adoption is low — the world is a long way from integrating AI into our daily lives. Although the public is “trying” AI, it’s only on a project basis. “If the software was as good as everyone says it was, the retention would be high”, Green told CNBC.

Another flaw: AI is not nearly as cost-effective nor as “transformative” as many experts had hoped, notes Jim Covello, head of Global Equity Research at Goldman Sachs. That was echoed by David Cahn, partner at venture capital giant Sequoia. He recently opined on the “$600B question” that is AI investment, and says that the gap between revenue expectations in the AI sector and actual revenue growth is widening. In his words: "the $125B hole is now a $500B hole”. In other words, it remains to be seen whether even major players such as Google, Microsoft, Apple and Meta will be able to make enough profit to make their substantial investments in this technology worth it.

Yet companies continue to pump money into AI development, hoping their bets will pay off.

Take, for example, voice AI platform SoundHoundAI, which IPO’d to great fanfare, quickly hitting $13.47 per share. It now stands at $4.89 per share, and the company reported operating losses last year of $28.52 million. But they aren’t an anomaly — this year alone, Wall Street expects $60 billion to be invested in AI, yet only $20 billion to be earned back.

So why would investors keep pouring money into what looks to be a sinkhole? Many of them would tell you that even though AI is a sinkhole right now, it won’t always be. They’d argue they’re playing the long game. Others are simply suffering from irrational exuberance. This is the official term for what happens whenever the market makes optimistic projections about a stock or an industry that simply isn’t rooted in economic reality.

According to research firm Gartner, the AI industry is now approaching the ‘Trough of Disillusionment’, a stage that immediately follows a period of hype (which is where we are now). If AI follows the same historic pattern as other industries, as we move through this trough, companies will either rise to the top by satisfying customers’ needs, or they’ll fail. The trick is that there’s no way to know exactly where to place your bets.

Investors who are pro-AI argue that even though the cost of the technology is expensive right now, it will come down over time, as it does with all tech cycles. And when it does, investors who have been playing the long game will reap the rewards. We just have to be patient.

But for how long? JP Morgan describes AI as a multi-year opportunity which is “only just starting to show up in corporate bottom lines”. In other words, we’re too early in the life of AI to see real benefits for investors in the near future. According to their calculations, the real potential for growth lies in the coming decades, and we shouldn’t worry about a bubble — comparatively, today’s AI stock prices are still far from tech bubble highs.

Their advice to investors looking to dabble in this brave new world? Build portfolios that include companies across the full spectrum of AI. For now, the best potential opportunities are in two areas:

The infrastructure that underpins AI, such as data centres, utilities and semiconductors. Stock picks in this category could include cloud computing companies, such as Microsoft or Alphabet, and semiconductor companies like Nvidia.

AI software and applications developed for specific industries, like customer service, health care, finance, and logistics. Stock picks in this category could include OpenAI, IBM and ElevenLabs, for example.

And once we wade through the ‘Trough of Disillusionment’, we’ll find ourselves on the ‘Slope of Enlightenment’. Once we hit it, we’ll know more about the cost-effectiveness of AI for the long haul, and we’ll have a better idea of how the sector will mature — in other words, we’ll know which AI bets are bigger, badder (and safer) than others.

Until we’re there, our best advice is something you’ve heard here before: Diversification is key. Don’t have all your eggs in one basket. Historically, investors have lost a lot of money during speculative technology waves, but you don’t have to. Acknowledge that with any AI-targeted investment, you’re chasing a trend. And know that during times of rapid development like this, picking a winner is hard, but picking a loser is, unfortunately, easy.

Oasis will reunite for a 2025 tour, but tickets are already sold out. After all, the 90s are hotter than ever.

Parental love really is the strongest. Sorry, the mums were right. The love for your kids makes your brain light up in ways that no other love does.

Votes for the best dressed royal are in.

102-year-old veteran becomes Britain’s oldest skydiver. If he doesn’t make you smile, nothing will.