In the financial world, there are a lot of self-proclaimed experts who have made a living pushing doom and gloom. The end is near. The sky is falling.
AI has now come into their crosshairs.
And the recently published MIT study showing that 95% of AI investments have not paid off has thrown gas on the AI naysayers’ fire. It also found that the Mag7 companies have spent $560 billion on AI over the last 18 months and generated $35 billion in incremental revenue, which some have gleefully noted would mean a rate of return of 0.6% (assuming a 10% profit margin).
They are also quick to liken the current environment to the dot-com bubble in the early years of this century, which famously crashed and burned to a large extent.
The recent correction/choppiness in the market is taken as evidence of a bubble (while there may be “some” truth to that, we’ve also enjoyed a long bull run for much of this year, and a correction is hardly unexpected).
But the doom-sayers are missing the point.
When a car company builds a new multi-billion-dollar factory, nobody expects it to recover the costs in 18 months.
When a pharmaceutical company spends billions of dollars over a decade or more developing a new drug, no one is looking at the rate of return in the R&D phase.
The AI megatrend has been gaining momentum for decades, but the release of ChatGPT in 2022 was an inflection point – a combination of breakthrough data science coinciding with new data processing technology to facilitate a quantum leap in what was possible, technically and financially, for businesses large and small.
In baseball terms, we are in the very early innings of the AI game. We have a lot of game left to play.
Bubble? Did Somebody Say Bubble?
As I mentioned above, many are quick to note the similarities to the dot-com bubble. And, yes, there are some similarities. But there are differences too. Very important differences.
On the “bubble” side, we have seen some stocks that have gotten a lot of lift from the AI talk. Perhaps none more than Nvidia. Are those multiples too high now? They are unquestionably high relative to history. But are they too high? Only time will tell. There are also plenty of companies that have latched onto AI the way they latch onto any much-hyped development, hoping their share price will bask in the glow of the latest megatrend. It should surprise no one that the SPAC IPOs that were popular in the 2019-2021 years before falling into disrepute as a way for founders to cash out and leave shareholders holding big losses, are back. So far this year, more than $22 billion has been committed to SPACs.
We also have some of the public AI platform/tool races for first-mover or early-mover advantage in many categories. That means for every category there may be 10, 20, or more contestants, when only six or so are likely to survive, and a handful of those will dominate. Some of the rest will get acquired and merged into bigger organizations, but many will limp along like zombies or fail completely.
That’s what happens with the development of breakthrough technologies. Remember Palm Pilots? Or Blackberries? AI does have some element of that, and if you were a big investor in Alta Vista before Google crushed it, you might be understandably wary of AI.
BUT … There’s a BIG Difference
The dot-com bubble brought a lot of so-called “vaporware” – great ideas that couldn’t be feasibly developed on a commercial scale because of software and hardware limitations, lack of capital, or lack of infrastructure.
Conditions are very different in 2025.
Computer processing power is DRAMATICALLY greater than it was 20 years ago. Memory is enormously faster and cheaper. 5G mobile technology is largely the norm, and the penetration of extremely powerful mobile devices is far beyond critical mass. Power generation capability is far greater and growing, using a wider range of technologies, many with falling cost structures, improving efficiency, or both.
And while there are many startups in the game, there are also enormous businesses – both tech companies like Nvidia, Amazon, Google (Alphabet), Facebook (Meta), and Microsoft as well as operating companies like manufacturers, banks, pharmaceuticals, airlines, etc. – with real businesses, real revenues, large R&D budgets and access to capital.
The development of generative AI technology was the capstone piece that brought it all together.
As I’ve written previously, this new AI megatrend will ultimately drive massive growth in productivity. I am 100% confident of this. And productivity gains create new wealth and improve standards of living.
But as the saying goes, “Rome wasn’t built in a day.” While we have already seen some benefits, the full effect of those AI gains will emerge over the coming decade and beyond.
A Warning
The other issue we need to consider is the social cost of AI. AI will change, create, and eliminate jobs. We already have a problem with income disparity, and with the “K”-shaped economy of the last several years, the rich have done well, but the rest have not.
It’s important to make sure some of the fruits of that productivity growth and wealth creation go to fund retraining and relocation. We didn’t do that adequately in the ‘90s and early aughts when we were dropping trade barriers and enabling more free trade. While those policies also improved productivity and many benefited, there were many who didn’t and whose jobs and industries were negatively affected, generating hard times as well as enduring anger and frustration that have deepened the divisions in our country with negative results. There were many winners but too many losers as well. We can and should do better.
That warning aside, the AI megatrend is real. Investors doubt it at their expense.
But not everything AI touches will turn to gold. Even savvy investors can benefit from the guidance of the kind of investment experts we have here at Cabot. Their research, analysis, insights, and recommendations help you find the winners and avoid the losers. That’s what we’ve been doing since 1970.
[author_ad]
