Is AI Even Profitable?

I love ChatGPT…

I love it so much I’m having a hard time remembering how I survived without it. 

I’ve saved countless hours using it to draft sample contracts, leases, and even stodgy business letters to my bank. 

I’ve also used it as a sounding board in creative brainstorming sessions. 

I play the saxophone. Just last week, I used it to transpose soprano sax sheet music to alto sax. 

I’m also a Texan living in Lima, Peru. When I’m struggling to find the right words in Spanish, I use ChatGPT as a translator.

It’s a phenomenal tool. 

And yet…

It hemorrhages cash. 

So, today, let’s ask the question few analysts dare to ask…

Is AI even profitable?

And if it’s not, what does that mean for folks who are investing in leading AI players today?

Colossal Costs

This year, the company behind ChatGPT, OpenAI, expects to bring in about $4 billion in revenues… and take a loss of at least $5 billion. 

Its expenses were more than double its revenues, in other words.

AI losses are showing up elsewhere, too. 

Microsoft is a major investor in OpenAI. 

Its Chief Financial Officer, Amy Hood, said the company lost about $1.5 billion last quarter on its investment. 

Ouch!

Yes, OpenAI’s revenues are growing as more people subscribe to the app. But so are the expenses. 

Training GPT-5 – the company’s upcoming AI model – is expected to cost as much as $2.5 billion. That’s 17.5 times more than the cost to train GPT-4. It’s nearly 400 times more expensive than the cost to train GPT-3.

Here’s the thing… 

OpenAI doesn’t have a choice. If it doesn’t spend the money, someone else will. 

Someone like Elon Musk.

Musk’s AI division – xAI – just spent about $4 billion on Nvidia chips for its gargantuan AI supercomputer, Colossus. 

And xAI is no closer to profitability than OpenAI. It’s running up billions of dollars in expenses and generating revenue on the order of just tens of millions. 

OpenAI and xAI are not the only companies forking out hundreds of billions of dollars on AI infrastructure.

Escalating “Arms Race” 

Amazon plans to invest over $100 billion in AI-capable data centers over the next decade. 

It’s also developing an AI supercomputer called the Ultracluster together with leading AI company Anthropic. 

The Ultracluster is expected to be up and running next year. 

Meanwhile, Mark Zuckerberg’s Meta has budgeted between $35 billion and $40 billion for AI infrastructure in 2024… with tens and tens of billions more spending on tap. 

Microsoft and Google parent company Alphabet are each expected to spend in the ballpark of $50 billion on AI in 2024… and even more in 2025. 

I could go on, but you get the idea. 

Tech firms are spending a gargantuan amount of money on AI development – with each iteration ten times more expensive than the one before.

And almost every major tech company feels obliged to escalate the AI “arms race” or risk getting left behind. 

I’m not a luddite or a technophobe. Let’s be clear on that. I love generative AI and use it almost daily. 

But what if the business model is fundamentally flawed and none of these companies make money?

It wouldn’t be the first time something like this happened. 

From Bright Hopes to Dark Fiber

Let’s go back to the 1990s, and the dawn of the Internet Age. 

Early on, it became obvious that slow dial-up internet speeds were an obstacle to growth. It also became obvious that rolling out high-speed internet would involve major capital spending on a previously unimaginable level.

Enter Global Crossing. 

In 1997, former investment banker Gary Winnick led a group of entrepreneurs to launch the company. It had a mission: Crisscross the world with a massive global network of undersea fiber optic cables. 

And it did!

By 2021, it had connected 200 major cities in 27 countries and laid an estimated 80,000 miles of cable. 

But it wasn’t the only company doing this. 

In the late 1990s, companies laid an estimated 80 to 90 million miles of fiber-optic cable. That’s enough to loop around the Earth about three and a half times… or get you a third of the way to the moon. 

These companies built out infrastructure because no one could afford to be left behind. 

We know what happened next. 

In March 2000, the dot-com bubble that had been inflating in the 1990s burst. 

And in 2002, only about 5% of the cables that had been laid were in use. 

The remaining “dark fiber” went unused for the better part of a decade until demand for video streaming finally caught up with the surplus capacity. 

Global Crossing – the poster child of tech infrastructure spending of the dot-com boom – filed for the fourth largest bankruptcy in history at the time. 

Next Phase of the AI Boom

Am I calling for the bankruptcy of today’s big spenders on AI infrastructure?

No.

These companies – the Microsofts, Amazons, and Metas of the world – are much bigger… and have much more diversified businesses… than Global Crossing.

But none of the companies that built the internet infrastructure of the 1990s turned a profit on their efforts. 

Instead, the infrastructure they left behind laid the foundation for some of the greatest money makers in history – Amazon, Google, Facebook, Netflix, and other tech darlings of the 2000s. 

Could we see something similar happen in AI?

Yes, we most certainly could. 

AI is fueling exponential progress and transforming the workforce to a degree not seen since the Industrial Revolution.

But the real winners may not be the companies spending hundreds of billions of dollars on AI infrastructure.

They’ll be the companies turning this tech into profits.

My team and I will be on the lookout for these AI profit makers as the next phase of the AI boom plays out. 

Already, we’re focusing on the “pick and shovel” plays that stand to profit from the AI arms race itself, regardless of who actually wins it. 

To find out more about the investments we already have in our Freeport Investor model portfolio – and to be first to hear about the next opportunity we find – click here.

To life, liberty, and the pursuit of wealth.