Listen to the audio version of this article (generated by AI).
Tom Yeung here, with your Sunday Digest.
Over the past several months, the most common question we’ve heard is this:
Is it too late to buy AI stocks?
You would be right to wonder. After all, most AI stock charts now look something like this…
Intel Corp. (INTC) stock
And when the air suddenly rushes out of hot stocks, it can really go out all at once. Below is a chart of Cisco Systems Inc. (CSCO) in the aftermath of the dot-com bubble.
Cisco Systems Inc. (CSCO) stock
But here’s the thing. People ignore technological changes at their own risk. The internet did end up changing the world. And if investors had avoided companies like Cisco and bought Amazon.com Inc. (AMZN) instead, they would have turned every $10,000 invested into $30 million through today.
The AI Revolution will do the same. It will create a handful of winning companies while leaving everyone else behind.
So, the trick is to figure out which of these hot firms still have more room to run…
Or better yet, find those that Wall Street has barely even discovered.
Our three InvestorPlace Senior Analysts have done just that. In a new presentation, Louis Navellier, Eric Fry, and Luke Lango join forces to talk about their “best in class” AI portfolio of 19 companies.
These companies span a wide range of industries. Some have already risen (and have more room to grow), while others are barely starting their upward journey.
But they all have one thing in common: They are the companies propelling the AI Revolution ahead, rather than the ones getting left behind by this new technology.
Today, I have been given special permission to reveal one of these top picks. And if you want to learn more about accessing their full list of 19 stocks, click here to sign up to watch a special event they are hosting on Wednesday, August 19, at 10 a.m. Eastern.
The Next Trillion Dollar Company
In June 2026, Nvidia Corp. (NVDA) CEO Jensen Huang got up on stage at a major technology trade show and introduced the next speaker:
“The next trillion-dollar company, ladies and gentlemen,” he said, beckoning at his fellow guest.
The person he pointed at was none other than Matt Murphy, CEO of a firm our three analysts have added to their AI Revolution Portfolio:
Marvell Technology Inc. (MRVL).
And there are good reasons to be so confident in Marvell’s future.
In short, Marvell is one of exactly two companies on Earth that can build custom AI accelerators and the connections between them. These are two “superpowers” that deserve their own explanations.
Superpower 1: The Custom Engine Shop
When most people hear “AI chip,” they think of Nvidia. That’s fair. Nvidia’s graphics processing units (GPUs) are the best general-purpose AI chips that money can buy.
But “general-purpose” is another way of saying it’s the “Ford F-150 truck” of the AI world. That’s because when it comes to cars, the 4-door pickup does everything quite dependably. It’s surprisingly fast, can haul a reasonable load, and is by far the most popular vehicle in America. That’s a lot like Nvidia’s flagship “Blackwell” AI chips.
Yet, no one expects a standard Ford F-150 to do very well in a drag race. These quarter-mile sprints favor stripped-down cars that are designed to do just two things:
- Go very fast in a straight line, and
- Don’t blow up.
That perfectly describes what AI “inferencing” often needs. These types of computing tasks follow the same known path repeatedly. That makes custom chips supremely good, since they are less flexible, but extremely fast on a set track.
That’s where Marvell’s custom chips come in. The company owns a vast library of chip parts, and can assemble them like a custom engine shop building a drag racer. Together with rival Broadcom Inc. (AVGO), these two firms control roughly 95% of the custom AI chip market.
Marvell’s engineering is already serving some famous customers. It is the force behind Amazon.com Inc.’s (AMZN) Trainium AI chips, where over a million have already been deployed. And Marvell is reportedly the engineering partner behind Microsoft Corp.’s (MSFT) next-generation Maia 300 processor, expected to be revealed in September. Alphabet Inc. (GOOGL), Apple Inc. (AAPL), and Meta Platforms Inc. (META) are also customers.
That’s why Marvell’s management expects its custom chip revenue to more than double next fiscal year, and to exceed $10 billion annually by fiscal 2029.
Superpower 2: The Digital Superhighways
Now, here’s the part most investors miss.
A modern AI datacenter is not one big chip. It is tens of thousands of chips working together. And if these chips cannot communicate efficiently, it would be like running a newsroom that forces reporters to file stories with carrier pigeons. It technically works… eventually.
That makes the connections between chips just as important as the chips themselves. And this is Marvell’s crown jewel.
Marvell controls roughly two-thirds of the market for optical digital signal processors (DSPs). These are specialized chips that convert electrical signals into pulses of laser light so that data can zip between servers over fiber-optic cables. They are far faster than copper wires, and can pack more data through tiny spaces, leaving more room to add more AI chips.
Perhaps the strongest endorsement came from Nvidia itself. Rather than fight Marvell, the “king of AI” decided to adopt the technology. In March 2026, Nvidia invested $2 billion in the company so it could put Marvell’s optical chips into its proprietary NVLink Fusion ecosystem.
This business is growing more than 70% per year.
Why Buy Now?
Marvell is already a well-established player of the AI Revolution. Management expects revenues to rise 40% this fiscal year to roughly $11.5 billion, then another 45% the following year to $16.5 billion. The company has beaten estimates and raised guidance four quarters in a row, and Wall Street’s profit estimates for fiscal 2028 have surged 52% in under a year.
These are fantastic numbers.
Yet, the stock has fallen sharply since June. Shares peaked at $330 shortly after Huang’s trillion-dollar introduction, then got dragged down to the low $200s by July’s tech selloff.
That provides investors with a second chance to jump in on Marvell’s shares. I expect the stock to grind higher over the next five years and have a bull case of over $350 if AI datacenter demand materializes as expected.
One warning: Marvell’s stock is volatile. Shares have gone from $60 to $330 and back to $220 in the past twelve months. Options traders are expecting a double-digit swing in either direction when Marvell reports earnings on August 27. Volatility is the price of admission here. It’s also what keeps handing patient investors terrific entry points, just like the one July’s selloff created.
Where the AI Revolution Will Head Next
Marvell checks every box we look for in an AI winner: real customers, accelerating revenues, and a moat only one other company on earth can cross.
But it’s just one of the companies that Louis, Eric, and Luke have identified. To learn more about the stocks they like right now… where they think the AI Revolution will head next… and how investors can properly position their portfolios to play it, I recommend you click here to reserve your spot for their special event on Wednesday, August 19, at 10 a.m. Eastern.
Louis will also be announcing a major change to his role at InvestorPlace. Sign up for the event now so you don’t miss it.
Until next week,
Thomas Yeung, CFA
Market Analyst, Investorplace
5 Stocks Our Experts Predict Could Double In the Next Year
By submitting your email, you'll also get a free pivot & flow membership. A free daily market overview. You can unsubscribe at any time.
