Image
Image

By Don Yocham, CFA

Posted: December 20, 2024

AI Eating Itself

Nothing goes up in a straight line forever.

Every trend, process, or dynamic has limits. They encounter negative feedback loops that send them running in the other direction.

Nature herself provides them in abundance.

Get hot, you sweat, and you cool down.

When rabbits breed too much, you’ll soon have a lot of foxes.

Don’t tell climate alarmists this, but higher temperatures lead to higher humidity and more cloud cover, putting real limits on temperature trends.

That’s not to say that things can’t go to extremes. But they don’t keep going on forever.

You see it in markets and the economy, too, especially when over-investment and speculation follow new innovation.

Railroads in the late 1800s, autos in the 1920s, and, of course, the internet had its reckoning at the turn of this century.

But AIs most immediate limit may not be too much capital.

That limit, as I will show you, comes from within.

AIs Negative Feedback Loop

AI models are trained on data.

That data doesn’t come from a database. It comes from articles, books, videos, and other human-generated content.

By reading, “watching,” and otherwise consuming all types of content, AI models learn to predict what a human would say or reason.

We’re seeing the results of that training with a flood of AI-generated content.

And it is a flood.

Analysts predict that up to 90% of all internet content could come from AI by next year. By 2030, we’re looking at 99%.

With predictions like these, it’s hard to see what will be left for humans to do.

But that’s where limits enter the scene.

AI models scour the internet for input. However, when that input is AI-generated—or when AI eats itself—the AI model will collapse.

Train an AI on AI-generated data and it compounds errors, draws poor conclusions from that data, and essentially just repeats itself.

Basically, it builds predictions off data that was itself a prediction, it learns too much of the same thing, and fails.

There are other limits too. Resources, including cooling equipment, building materials, networking, and the power to run the models all act as constraints to the growth of AI.

Now, don’t get me wrong. AI will have a massive impact just like trains, cars, and the internet had massive, positive impacts on economic growth.

But it’s important to always keep in mind that most business models emerging around highly disruptive innovations fail. They either miss the point or invest too heavily to weather the inevitable correction.

Only a few make it through to emerge as winners on the other side.

Which means you need to work really hard to focus your capital on those companies best situated for the long run.

In the case of AI, Nvidia Corp. (NASDAQ: NVDA) looks like a clear long-term winner.

Expensive, but Worth It

The market wants Nvidia chips.

And it’s willing to wait.

That’s because Nvidia has spent well over a decade debugging software and driver issues for its CUDA software programming system. This gives it a critical edge as it will take years for its competitors to work through the same issues.

It also works with any cloud platform, giving customers crucial flexibility, especially relative to Amazon and Google custom chips, which only work on their respective cloud platforms.

Nvidia’s market share, platform stability, and developer toolkit make it the preferred vendor for developers. An entrenchment that actually creates a positive feedback loop for its future growth.

Finally, Nvidia’s chips outperform the competition.

Corporate investment spending on AI is just getting started. And it’s already massive.

Microsoft, Google, Meta, and Amazon will each invest around $40 to $50 billion in data centers over the next year to build out their capacity to deliver AI solutions.

As Eric Schmidt, former Google CEO, recently stated, “If $300 billion is all going to go to Nvidia, you know what to do in the stock market.”

As I show in this Capital InFocus report on Nvidia, the stock is rich but worth it.

There are no comparable alternatives to Nvidia chips. The stock won’t go up in a straight line forever, but from where we stand now, it has the best chance of weathering the inevitable AI correction to emerge a winner on the other side.

Think Free. Be Free.

Don Yocham, CFA

Managing Editor of The Capital List

Related ARTICLES:

Image

AI Eating Itself


By Don Yocham

Posted: December 20, 2024

Nothing goes up in a straight line forever. Every trend, process, or dynamic has limits. They encounter negative feedback loops that send them running...

Image

Tesla: The Ultimate Trump Trade


By Don Yocham

Posted: December 16, 2024

“Trump trades” are all the rage right now. But that trade isn’t really “this stock” or “that stock.” It’s the whole market. Tax cuts,...

Image

Markets Rising for a Reason


By Don Yocham

Posted: December 13, 2024

Yesterday, I described some factors that could keep this market moving. This economy has so much going on under the hood that even today’s...

Image

Gensliger’s Out


By Don Yocham

Posted: December 5, 2024

I laid out in October the disaster unfolding at Intel Corp (NASDAQ: INTC). The company began losing its competitive edge in 2018 when it...

Image

The Elemental Energy Link


By Don Yocham

Posted: November 21, 2024

Every three days, a new data center opens somewhere in the world—with many classified as hyperscale. These data centers consume massive amounts of energy....

Image

Holding Capital to Account


By Don Yocham

Posted: November 1, 2024

To paraphrase Trump, Big Tech has spent “bigly” on AI data. Microsoft Corp. (NASDAQ: MSFT) invested $20 billion in AI during the latest quarter....

Image

Energy To Ride Out the Storm


By Don Yocham

Posted: October 30, 2024

We will experience a monumental election in just a few days. Despite a strong economy (at least according to the stats), low unemployment, and...

Image

Coming Down the Mountain


By Don Yocham

Posted: October 9, 2024

Throughout the commercial computer era, which began roughly in 1970, Intel Corp. (NASDAQ: INTC) remained a leader among chip makers. You could say it...

Image

The Next AI Contender


By Don Yocham

Posted: October 2, 2024

The race to build out computing power to meet booming AI demand is currently underway. The best way to gauge that expected growth is...

Image

The Price of Profits


By Don Yocham

Posted: September 27, 2024

In a few of my recent notes, I’ve discussed how to think about a company’s market value. Capital forms the foundation of that value....

FREE Newsletters:

"*" indicates required fields