In partnership with
The AI stock market rally shows no signs of stopping.
Nvidia's market cap recently hit $5 trillion. Barely 2 years ago, it had hit $2 trillion for the first time, and the market started calling it “bubbly”…
The bubble-callers have been wrong for so long that people are starting to ask… what if AI is real? What if AI isn’t a bubble?
The question isn't whether AI technology is revolutionary. It quite obviously is. The question is whether the current valuations reflect reality or hype. Whether you should trim your positions, or double down. And, whether or not this ends like the dot-com boom (transformative technology, but with a brutal correction in the middle).
Today, we'll cover:
What the latest data shows
If AI is in a bubble or not
The similarities and differences between AI and the dot-com era
How to position yourself to win either way
Let's dive in.
But first, if you want to future-proof your income streams from AI disruption, consider joining the Wealth Potion Academy.
The Academy features over 20+ lessons across four modules, or “stats”: Income, Savings, Investing, and Time.
And the Income module is crafted specifically for ensuring your skills, your career, and your business are resilient to AI disruption.
7 of the 24 lessons are currently accessible and once they are all published, early-bird pricing will be unavailable. So jump in early if you’re ready to get started.
Now back to the article.
Why This Isn't the Dot-Com Bubble (Yet)
The dot-com comparison gets thrown around constantly. I believe this is either lazy analysis, or clickbait, or both.
In 1999, the malinvestment was much more blatant.
Pets.com burned through $300 million selling dog food at a loss. Webvan raised $800 million to deliver groceries in 10 cities before collapsing. eToys went from a $7.8 billion valuation to bankruptcy in 18 months.
These, quite frankly, weren't real businesses.
Nvidia, by contrast, just posted $60 billion in annual revenue with 73% gross margins. This is a very real business. Real companies are paying Nvidia real money for real products and services.
The infrastructure layer of AI (the picks and shovels of this AI gold rush) is already profitable. Massively profitable.
So where is the froth?
What's Real vs. What's Not
As a quick aside, the term "froth" in investing refers to when prices become decoupled from value due to hype or euphoria. So when markets are "frothy", that means to beware of the hype.
But in order to understand what’s not real, we first have to look what is real.
Source: isaiprofitable.com - note the major exception!
The Infrastructure Layer is Real
Large language models work. This is obvious to anyone who has used ChatGPT or Claude. GPT-4, Claude, and open-source alternatives like Llama are solving actual problems for actual businesses. Software engineers are 30% more productive with GitHub Copilot. Customer service teams are handling 3x more tickets with AI assistants. Marketing teams are creating content at scale.
Put simply, AI is creating real productivity gains for its users.
Enterprise Adoption is Real
Fortune 500 companies are rebuilding entire workflows around AI. Microsoft is embedding Copilot across Office. Salesforce is adding AI to every product. Even very conservative industries like banking and healthcare are running AI pilots.
When enterprises start budgeting $100,000+ annually for AI tools, that creates sustainable revenue streams for the companies building them. So the comparison to the internet is apt: virtually every business today is an internet company, because they in some way have incorporated the internet into their operations. AI is following a similar trajectory.
Inference Costs Are Really Dropping
Inference costs are the costs incurred by the user to run the AI model, and running AI models is getting cheaper every single quarter. What cost $20 to process a year ago now costs $2. That makes AI viable for use cases that weren't economical before — like replacing human customer service at scale.
There is little-to-no evidence that this trend of decreasing costs is going to reverse, stop, or slow down any time soon.
Now, let's get into what isn't so real, yet.
The Application Layer is Frothy
Half the AI startups in Silicon Valley are building features, not products. "AI-powered" email assistants. "Smart" note-taking apps. Chatbots for restaurant reservations.
We're already seeing LLM providers like OpenAI and Claude launch features that are basically wiping out entire startups.
AI startup founders need to have an honest conversation with themselves (and their teams) on whether or not they are building a feature that OpenAI will release in the foreseeable future.
Unprofitable Model Trainers are Frothy
OpenAI is reportedly losing money on every ChatGPT interaction. Anthropic burned through hundreds of millions training Claude. Most AI labs are still 2-3 years away from profitability, if they ever get there.
This reveals some similarities to the dot-com bubble. Many are pointing at this alone and calling it blatant malinvestment. But remember: Amazon was unprofitable in late 2001, 7 whole years after its founding. And looking back, that was a very smart business decision by Jeff Bezos to ignore profitability and pursue growth at all costs. Amazon is currently the largest company in the world by revenue.
Everything Getting an "AI" Label… Froth
SaaS companies are rebranding basic automation as "AI-powered." Marketing agencies are calling themselves "AI consultancies." Even traditional software companies are adding "AI" to their pitch decks to boost valuations.
When everyone claims to be an AI company, most of them aren't. Once again, this has some similarities to the dot-com bubble. But this time around, because there are real use cases being developed and used across the marketplace, these hollow rebrands are easier to spot.
The Nvidia Exception: Why the Pickaxe Sellers Win
It’s worth discussing what makes Nvidia different.
Why is Nvidia massively profitable while OpenAI, Anthropic, etc. are not?
Following the California Gold Rush of the mid-1800’s, most of the gold miners went broke. But the people selling picks and shovels got rich. This became a euphemism in investing: in a gold rush, sell picks and shovels.
Nvidia is the AI “picks and shovels” seller. Every AI company needs their chips. Whether OpenAI succeeds or fails, they still need H100s to train models. Whether your favorite AI startup gets acquired or shuts down, the underlying compute was purchased.
Nvidia is currently outperforming because it is a bet on the entire digital transformation of the economy. Cloud computing, gaming, autonomous vehicles, data centers. AI is just the latest (and biggest) driver of chip demand.
The math is pretty simple: if AI continues growing, Nvidia prints money. If AI growth slows, they still dominate the data center chip market.
Compare that to OpenAI, which is burning billions while competing with Google, Microsoft, Meta, and Anthropic. Even if they "win" the AI race, their profit margins will get compressed by competition. Meanwhile, all those competitors are buying Nvidia chips.
How to Position Your Portfolio
None of this is specific financial advice, but principles you should be thinking about - especially if your portfolio is AI-heavy:
Don't try to time the exit. Nobody rings the bell at the top. The companies with real revenue and sustainable business models might keep running for years. The unprofitable feature plays might crash next quarter. It is extremely difficult to predict which is which.
Trim positions as needed. If AI stocks are growing so fast that they are becoming an uncomfortable proportion of your portfolio, it is okay to take some profits. You can always buy back in if you're wrong. Just don't break rule #1 above - don't try to time the market.
Distinguish between infrastructure and applications. Nvidia, Microsoft, and Google have diversified revenue streams and sustainable AI businesses. Random AI startups do not.
Use dollar-cost averaging to reduce exposure. Instead of selling everything at once, trim a small portion of your AI positions every month for six months, for example. This smooths out the timing risk.
Stay confident in what you actually own. If you bought Nvidia at $200 and it's now at $800, taking some profits isn't "missing out." You already won. Banking some gains lets you sleep at night while staying exposed to further upside.
Or, if you still have conviction that there is further upside, then keep on holding. This is not financial advice. But hopefully, it is a helpful framework if you have been considering a trim, but have been holding out.
The Bottom Line
If you were looking for a hard yes or no answer to the AI bubble question, I'm sorry to disappoint you. But the honest answer is that nobody knows for sure.
That said, the market is the ultimate source of truth in investing. The fact that prices keep going up tells us that more people believe it is real than not.
And in a way, the AI bubble question misses the point... AI isn't just one thing. It is a paradigm shift that is slowly but surely changing the way we work, the way we do business, the way we use the internet, and more. By its very nature, it is going to produce very polarizing opinions about it.
Here are some things we do know:
The infrastructure is real. Nvidia's revenue is real. Enterprise AI adoption is real. The productivity gains are measurable and sustainable.
But if we look at who is buying from Nvidia, we notice something else...
The application layer is causing the froth. Too many companies chasing the same opportunities with unsustainable business models and venture-funded burn rates. Some of these companies will win, and they will win big. But most of them will not.
So trim the froth if it's making you uncomfortable, keep the foundation so you don't make a move you regret, and don't try to time the exit perfectly.
AI is real. And it is just getting started. But that doesn't mean every AI stock goes up forever.
Ready to build wealth regardless of market bubbles? Join the Wealth Potion Academy and learn the frameworks that work in any market cycle.
To your prosperity,
Brandon @ Wealth Potion
Sound familiar?
Over 4 million people have had the same lightbulb moment.
Morning Brew is a free daily newsletter that breaks down what's happening in business, finance, and tech — clearly, quickly, and with enough personality to make it the best email in your inbox.
No yelling. No filler. Just the news, finally making sense.