AI Investors Face A Range of Risks

 

 

“The basic principle of venture capital is to put a dollar in each of 10 companies, accept that three will go to zero, one or two to $10 or more, and the rest kind of meh. The winner more than makes up for the losers, but you spread your money around in the hope of securing ten-baggers.

 

Investing in artificial intelligence increasingly has the same mindset -- but without the diversification. This leaves investors exposed to all the many risks as they chase one big bet: artificial general intelligence, an AI that can match or surpass humans. This isn't a chatbot, but a truly capable alternative to the human brain: Think Terminator, Hal, Blade Runner.

 

If what's known as AGI ever works, it would deliver massive societal change, as well as potentially huge productivity gains and, barring state seizure, profits on a literally science-fiction scale.

 

This is the better-than-ten-bagger bet that AI luminaries talk up as they set out plans for trillions of dollars to be sunk into data centers.

 

Along the way, there's "agentic AI" and other propositions that could still produce decent productivity gains and make lots of money if widely adopted. The problem: The entry cost is increasingly high because of the race to AGI, and even the prospects for take-up of the less exciting propositions remain uncertain.

 

Excluded from the VC world, ordinary investors are mostly invested in AI via funds that have a slice of their money in private companies such as OpenAI, or through Big Tech companies that are dedicating more of their cash piles to AI research and holdings in private AI firms.

 

Here are the risks:

 

It never works.

 

Clearly, if AGI proves impossible, those betting the farm-converted-into-a-data-center on it working will lose. Will it work? I have no idea. But there are good reasons to think that simply throwing more computing power at the current models won't do it, and every past AI boom has come with similar excitement about AGI.

 

AI pioneer Marvin Minsky told Life magazine that true AGI was three to eight years away -- in 1970. Less widely known is that the magazine canvassed skeptics who cautioned that AGI was more likely to take 15 years. It's always been the technology of the future, and it might stay that way.

 

It takes too long.

 

Maybe we do get AGI eventually. But how long will investors stay hopeful? The chips being bought today to fill the data centers will be redundant in four or five years, so need to generate significant revenue quickly.

 

It's highly unlikely they can be paid for with AGI, so they need either investors to keep on funding heavy losses, or customers to be willing to pay for other AI applications created along the way: video editing, fake digital friends, short online answers to replace search.

 

It costs too much.

 

Here we get to the problem of the other AI applications. Staff at OpenAI, creator of ChatGPT, just sold shares at a valuation of $500 billion. It is on track to meet its forecast of $13 billion in sales this year, almost all from the chatbot, valuing it at 38 times revenue. That multiple is a little more than dot-com poster child Cisco Systems at the peak of the tech bubble in 2000.

 

Even if everything goes well, the high entry cost lowers the return prospects, barring AGI. Plus, investors have to continue to believe it will go right and keep funding losses until it does.

 

You pick a loser.

 

Back in the browser wars of the late 1990s, Netscape fought Microsoft's Internet Explorer for dominance of the web interface. Netscape had the better product in my view, but it didn't matter because the eventual winner was. . .neither of them.

 

Google, now Alphabet, came up with Chrome and dominates use. All the big tech companies are competing to develop AI, along with startups in Europe and China. If you could somehow invest in all of them, you could hope to get the VC-type returns, even if all but one go to zero. But the winner might be someone else entirely.

 

Investors have already repeatedly missed when trying to pick winners among smaller AI stocks.

 

Super Micro Computer shares are worth a bit more than one-third of last year's peak after a wild run-up. SoundHound AI shares lost three-quarters of their value this spring after rising more than 12-fold in a year, before a rebound. C3.AI shares are worth less than half their 2023 high, and have lost almost 90% of where they briefly traded at after the 2020 IPO.

 

Competition rises.

 

Investors are betting that AI is a winner-takes-most market with fat profit margins.

 

It could be that usage speeds design improvements, creating a network effect where more customers mean the product gets better, so attracts more customers. Or it could be, as today, that there are a ton of different chatbot companies offering very similar products -- and margins are driven down both by competition and by the need to keep up heavy spending on research.

 

It's too easy.

 

AI stocks wobbled earlier this year when China's DeepSeek AI released a chatbot that was easier to train.

 

Academics in China just released a paper on their "SpikingBrain" model that claims a new approach allows powerful AI development on low-cost microchips.

 

If someone comes up with a way to provide cheap AI, it might speed up the path to AGI -- but it will hurt chip maker Nvidia and companies that filled data centers with its expensive, high-powered chips.

 

The economy suffers.

 

The scale of data-center expansion is stretching the ability of the U.S. to provide enough electricity, backup generators and other equipment, as it is now measured most effectively in percentages of GDP. The spending helps spread money from AI investment more broadly across the economy. But it also makes it harder for old industries that have to compete for power and equipment with companies flush with the gusher of AI cash.

 

With the U.S. operating at or close to full capacity, this means either inflation or a reallocation of resources -- both painful options.

 

If you think an AGI would generate multitrillion-dollar profits, even a tiny chance that it happens is worth a lot. Investors merely hoping for profits from intermediate AI products should be bothered about financing a sector increasingly priced for AGI.” [1]

 

1. Streetwise: AI Investors Face A Range of Risks. Mackintosh, James.  Wall Street Journal, Eastern edition; New York, N.Y.. 06 Oct 2025: B1.