Is There an AI Bubble? AI Company Losses Are Massive—and Growing

As generative AI companies race to develop more advanced models, a pressing question emerges: Are we in an AI bubble?

Long-time AI watchers are increasingly asking the question, especially as new reports emerge about OpenAI’s staggering losses. According to The Information, the company’s on track to lose up to $5 billion in 2024.

So, are we in a bubble? Is AI over-invested and overhyped?

I got the answers from Marketing AI Institute founder and CEO Paul Roetzer on Episode 107 of The Artificial Intelligence Show.

OpenAI's eye-watering losses

According to an analysis by The Information, OpenAI could be facing losses of up to $5 billion this year. This projection is based on:

• AI training and inference costs estimated at $7 billion for 2024
• $1.5 billion in staffing costs for about 1,500 employees
• Total operating costs potentially reaching $8.5 billion
• Projected revenue between $3-3.5 billion

Other companies like Anthropic are also burning through cash, with estimated losses of $2.7 billion this year and significantly less revenue than OpenAI.

But there may be more to the story than meets the eye.

The bigger picture: a calculated risk

While these numbers might seem alarming, Roetzer offers a different perspective:

“As crazy as it sounds, losing $5 billion or more in a year is insignificant in the grand scheme of what they’re pursuing and the market potential they’re pursuing,” says Roetzer. “$5 billion is nothing.”

Roetzer points out that major tech companies like NVIDIA and Google have R&D budgets that dwarf these losses. And the key is to understand the strategic thinking behind these massive investments.

Roetzer points to a recent post by Doug Clinton of Deepwater Asset Management, where he frames the AI investment strategy as a modern version of “Pascal's wager.” 

Pascal’s wager is a philosophical argument posed by philosopher and mathematician Blaise Pascal in the 17th century. The original argument essentially broke down belief in the existence of God into a rational probabilistic argument, not an emotional or belief-driven one.

(The result being: Even the most skeptical non-believer should logically just choose to live as though God exists, since there’s no real downside to being wrong.)

Clinton makes the same argument about AI, structuring the thinking in the following manner:

1. If AI is profound and you invest heavily: You capture hundreds of billions in incremental annual revenue.
2. If AI is overblown but you invest: You waste money on infrastructure that can be repurposed.
3. If AI is profound but you don't invest: You lose significant ground to competition.
4. If AI is overblown and you don't invest: You save money but miss out on potential trillion-dollar markets.

Basically, there are infinite downsides to not betting on AI. You don’t gain much by being right. You stand to lose a lot if you’re wrong.

This framework explains why tech giants are making such enormous bets on AI.

Tech leaders double down

Recent statements from industry leaders reinforce this strategic approach, says Roetzer. He points to a post from David Cahn, a partner at Sequoia, that breaks down the game theory at work here even further.

In other words…

The “bubble” behavior is actually more likely highly rational behavior from the major players in AI.

What this means for businesses

While the losses and investments in AI might seem staggering, they represent a calculated bet on a potentially transformative technology. As Roetzer puts it, "At the end of the day, it looks like the models are going to keep getting bigger and smarter."

That makes sitting on the sidelines exceptionally risky. Companies that wait to see if AI will be "profound" may find themselves left behind.

For now, all signs point to continued growth and investment in AI. The question for businesses isn't whether or not to engage with AI, but how quickly and strategically they can do so.