Multibillionaire and founder of Amazon (AMZN), Jeff Bezos, had a few things to say about artificial intelligence (“AI”) over the weekend. Speaking at Italian Tech Week, he urged investors to exert some caution.
When people get very excited, as they are today, about artificial intelligence, for example… every experiment gets funded [and] every company gets funded: the good ideas and the bad ideas. And investors have a hard time in the middle of this excitement, distinguishing between the good ideas and bad ideas.
He even used the dreaded B-word – bubble – to describe what we’re seeing with AI research and investment these days.
Now, keep in mind that Amazon is one of the world’s largest and most generous backers of AI. Last year, its capital expenditures came to $83 billion, most of which went toward AI. This year, the company is on track to spend $100 billion. And 2026’s tally, I imagine, will be as big or bigger.
So clearly, Amazon believes in AI.
As does Bezos himself, for the record. Even though he’s no longer CEO, he still owns 8% of the company and serves as its executive chairman. If he didn’t want Amazon to invest a fortune in AI, it wouldn’t.
That’s why he also told Italian Tech Week attendees that the technology itself remains a worthwhile pursuit:
But it doesn’t mean that anything that is happening isn’t real. AI is real. And it is going to change every industry.
He likened our current AI environment to the biotech bubble in the 1990s – which did indeed result in numerous companies crashing and burning. However, Bezos pointed out, “we did get a couple of lifesaving drugs” out of the ordeal. And the companies that survived went on to even bigger discoveries and profits.
Bezos is right, of course. Every new, exciting technology inevitably comes with overbuilding. It happened with railroads in Great Britain in the 1800s. It happened with the Internet in the late ’90s. Even in my industry – commercial real estate – there was massive overbuilding of retail in the 1990s.
Each bubble eventually deflated, but it didn’t mean the endeavor wasn’t worthwhile.
But it does mean you’ll want to be cautious about which AI investments you throw money after, now more than ever.
Widening Room for AI-Driven Financial Failure
Amazon itself is an excellent example of the few that succeed. Founded in 1994, it provided a very simple service (selling books) using what was then cutting-edge technology.
It wasn’t the earliest in on the e-commerce idea. Boston Computer Exchange actually launched its online marketplace in 1982. In which case, Amazon was over a decade “late” in entering the dot-com game.
It also wasn’t the biggest player back then. AOL, for instance, hit a sweet $200 billion market cap by the close of the century – whereas Amazon was worth less than $6 billion.
Yet we all know what happened to AOL afterward. So, the investing answer isn’t about size alone. In fact, when run by imprudent management, the bigger entities fall all the harder.
(Speaking of that, stay tuned for my article tomorrow about how scale has its advantages and disadvantages.)
Unfortunately, there is quite a bit of imprudence going on right now. Fortune 500 Digest noted in its Saturday e-mail how:
AI startups are under a lot of pressure to prove they can scale quickly. That’s because some truly have grown at an unprecedented rate, thanks to market conditions that VC firm Andreessen Horowitz has dubbed “the great expansion.”
The hottest startup metric for flaunting fast growth is ARR (annual recurring revenue). But some founders are getting creative about how they’re accounting for ARR as they attempt to raise new rounds of financing.
One VC told Fortune‘s Allie Garfinkle that what he’s seeing more closely resembles “vibe revenue” than truly recurring deals.”
Just as with the dot-com bubble (and the biotech bubble before that… and the housing market bubble after), there’s a lot of hype out there right now. That means there’s a lot of opportunity for FOMO (the fear of missing out).
And that opens the door to long-lasting financial regrets that I, for one, want nothing to do with.
Never Bet Too Much on the Unproven – Even if It Is AI
Right now, my team and I are researching artificial intelligence and the data centers that enable it.
You can read last week’s article on the latter here. But know this: There are a lot of pitfalls to navigate.
The Wall Street Journal wrote recently that it’s “time to stress-test everything,” including data centers. Writer Andy Kessler notes how:
Microsoft invested in OpenAI, which received credits for data-center computing, boosting Microsoft’s revenue. Same for Google and Anthropic. Oracle announced a $300 billion deal for data centers with OpenAI – money it doesn’t have. This was followed by Nvidia’s $100 billion investment (cash on hand is $56 billion) in OpenAI, which will then turn around and buy Nvidia chips.
Meta is borrowing $26 billion to fund a Louisiana data center and signing engineers to NBA-like contracts. But where are the customers? AI demand needs to grow like weeds on steroids to justify these deals, or else they’ll implode.
I’ll also point to Fermi (FRMI), one of the companies I mentioned in last week’s data-center evaluation. It’s a brand-new data-center real estate investment trust startup that just held its IPO on October 1.
Fermi soared more than 35% that day. And it traded as high as $32 before ending the week at $28.60. That gives it an implied market value of around $17 billion…
Despite having no operating assets yet and therefore no revenue. That won’t happen until next year, if everything goes as planned.
Thinking that it’s worth $17 billion at this stage is ludicrous.
Yes, Fermi intrigues me. It’s officially on my watchlist, and I do believe it has the potential to earn its current valuation eventually.
But that’s years from now. I’m not recommending it until it proves itself.
In the meantime, there are many data centers and associated companies that already have. They have established and growing customer bases. Their revenue is increasing. And their balance sheets are well-managed.
Those are the companies that should survive when the AI bubble does eventually deflate. They’re also the ones that, like Amazon 25 years ago, should go on to dominate out of the aftermath.
Let’s refuse to settle for anything less.
Regards,
Brad Thomas
Editor, Wide Moat Daily
PS: Stay tuned for the Wide Moat Show on Thursday, where my co-host, Nick Ward, and I will be talking about the hottest tech stocks.
|
