If I asked you to name the biggest winners of the AI boom, I’d doubt Qualcomm (QCOM) would be near the top of your list.

Nvidia (NVDA)? Absolutely.

Broadcom (AVGO)? Increasingly.

Qualcomm? Most investors still associate the company with smartphones.

That may be about to change.

This week, Qualcomm announced a major new partnership with Amazon (AMZN) to develop multiple generations of custom chips for Amazon Web Services’ massive AI data centers.

The companies will work together on chips designed primarily for AI inference, while also developing advanced optical technology capable of moving enormous amounts of data between chips and servers.

The potential numbers caught Wall Street’s attention immediately.

Amazon’s agreement is tied to as much as $60 billion of business with Qualcomm. Amazon also received warrants allowing it to purchase up to 25 million QCOM shares for $161.26 each, potentially giving it a roughly $4 billion stake.

Qualcomm shares jumped on the news.

But I think investors should look beyond yesterday’s move because this deal tells us something much bigger about where the AI revolution is heading.

AI Chip Boom 2.0

The first phase of the AI infrastructure boom was relatively straightforward.

Buy Nvidia GPUs.

The demand became so overwhelming that Nvidia went from a gaming-chip company to the backbone of the AI revolution. Its stock soared, its valuation exploded, and Jensen Huang became one of the most recognizable CEOs in the world.

When ChatGPT was released to the world at the end of November 2022, Nvidia was worth about $415 billion. Within three years it was the largest company on the planet with a valuation over $5 trillion.

There’s no question that Nvidia remains the undisputed king of AI chips today. But when an industry becomes this enormous, competitors inevitably start coming after the throne.

More importantly, Nvidia’s largest customers are looking for alternatives.

Amazon, Microsoft (MSFT), Meta (META), and Google (GOOG) are spending hundreds of billions of dollars building AI infrastructure. At that scale, even modest improvements in cost, performance, or energy efficiency can save billions. That’s why these companies increasingly want custom chips built specifically for their own AI workloads.

And that’s creating opportunities for companies most investors didn’t originally associate with the AI boom.

Qualcomm now expects its data-center business to generate approximately $5 billion in revenue in fiscal 2027 and $15 billion by fiscal 2029.

Amazon joins other major technology companies working with Qualcomm as the chipmaker attempts to diversify away from its longtime dependence on smartphones.

That’s a major transformation for a company that’s historically lived in your pocket.

Everyone Wants Custom Silicon

Qualcomm isn’t alone.

Broadcom has become one of the biggest beneficiaries of custom AI chips, working with hyperscalers to design specialized silicon.

Marvell Technology (MRVL) recently struck a major agreement with Google to help develop custom AI chips. The deal could ultimately generate as much as $120 billion in revenue through fiscal 2033 if performance targets are reached.

Then there’s Advanced Micro Devices (AMD), which continues challenging Nvidia in AI accelerators, and Arm Holdings (ARM), whose architecture increasingly sits underneath CPUs and other processors used throughout data centers.

Even Intel (INTC), after years of falling behind, could potentially benefit if its foundry operation becomes an important manufacturing alternative for companies designing custom chips.

This is what happens when a megatrend gets big enough.

The opportunity begins to expand far beyond the first wave of winners.

Beyond the Chips

This is where investors should be looking next.

AI data centers don’t simply need processors. They need networking equipment capable of moving enormous amounts of information between them. They need high-bandwidth memory. They need optical connectivity. They need advanced semiconductor manufacturing and packaging.

Qualcomm’s Amazon announcement specifically includes optical connectivity reaching 1.6 terabits per second. That’s important because Qualcomm spent approximately $2.4 billion acquiring Alphawave earlier this year to expand its data-center connectivity capabilities.

So, the next generation of winners could extend well beyond QCOM, AVGO, AMD and MRVL.

Companies like Arista Networks (ANET) can benefit from exploding networking requirements. Micron Technology (MU) sits at the center of the memory boom. And semiconductor-equipment leaders such as Applied Materials (AMAT), Lam Research (LRCX), and KLA (KLAC) can benefit regardless of whose logo ultimately appears on the chip.

That’s the picks-and-shovels opportunity I find particularly attractive.

You don’t necessarily need to predict who eventually takes market share from Nvidia. You can invest in the companies providing the equipment, networking, memory, and connectivity required as the entire AI semiconductor market expands.

The Amazon-Qualcomm deal isn’t evidence that Nvidia’s AI dominance is ending.

It’s evidence that the opportunity Nvidia helped create has become too big for one company.

And for investors searching for the next wave of AI winners, that’s very good news.