This Chip Designer Is Cracking the AI Code

Editor's note: When an established company ventures into a new market and fails, it often retreats to its comfort zone. But as Joel Litman from our corporate affiliate Altimetry explains, one major semiconductor designer is getting back on a horse that bucked it years ago. This company's early data-center projects flopped, and investors forgot about it. But now it has returned to the market in a big way...


Qualcomm (QCOM) spent years trying to convince investors that it could do more than make smartphones smarter...

The company dominated the cellphone chip market for many years... after the iPhone ushered in a new era for mobile devices in 2007.

Then, in the late 2010s, Qualcomm shifted more of its operations into cars, PCs, and physical objects like sensors, appliances, wearables, and industrial equipment.

This was right around the time investor attention started shifting to AI. But Qualcomm couldn't keep up. Its initial AI efforts proved unsuccessful, and it began scaling back its data-center operations in 2018.

Revenue peaked at $44 billion in its fiscal 2022 and then slipped to $36 billion in 2023, as global smartphone shipments and sales fell by about 4%.

Qualcomm survived the slump, with revenue recovering to $44.3 billion in fiscal 2025. But the company has been treading water since then.

Today, I'll explain how Qualcomm is reentering the data-center market... and why investors have yet to price in a return to stronger profitability in the age of AI.

Chips Power AI... AI Powers Profits

Qualcomm's data-center push is no longer a side project... 

Smartphones run on a tight power supply, so their batteries need to last. Data centers now face a similar constraint... on a much larger scale.

At its June 24 Investor Day, the company unveiled its road map for the AI and data-center markets. This included plans to launch the Dragonfly C1000, a high-efficiency central processing unit ("CPU") designed to handle workloads from data centers and agentic-AI models.

Qualcomm is also building a suite of data-center products designed to maximize performance per watt and lower total costs over the life of chips.

According to Qualcomm, Meta Platforms (META) has agreed to use the Dragonfly C1000, and future iterations of the CPU, in its AI infrastructure starting in 2028.

AI power demand keeps growing... making energy one of the biggest bottlenecks for hyperscale operators.

Qualcomm is bringing its expertise in chip design to that environment to increase efficiency.

Its data-center expansion should produce billions of dollars in revenue starting next year... 

During its Investor Day, Qualcomm raised its fiscal 2029 non-handset revenue target to $40 billion, roughly twice its prior goal. At least $15 billion of that revenue is expected to come from data centers.

This foray into data centers is a major shift for a company that most investors still associate with cellphones.

And that seems to be what the market is worried about...

Qualcomm's smartphone revenue came in at $5.1 billion in the third quarter... down 20% year over year.

This is exactly why the data-center pivot matters.

Despite the record demand in this space, investors aren't optimistic about Qualcomm's long-term performance... 

We can see this through the Embedded Expectations Analysis ("EEA") framework, one of the tools we use at Altimetry.

The EEA starts by looking at a company's current stock price. From there, we can calculate what the market expects from the company's future cash flows. We then compare that with our own cash-flow projections.

In short, it tells us how well a company has to perform in the future to be worth what the market is paying for it today.

Over the past decade, Qualcomm's Uniform return on assets ("ROA") was often above 20%. At Altimetry, we analyze earnings with Uniform Accounting to avoid the distortions of traditional accounting methods. A 20% Uniform ROA means the company generated big profits through its core chip franchises.

When the handset downturn hit, the company's revenue fell, and investor confidence collapsed. After three rough years, the market appears to be treating the 20% Uniform ROA as a hard cap.

Take a look...

The market expects the post-2022 version of Qualcomm to persist. It expects the company's ROA to stay roughly flat, near those levels, after this year, even as its revenue mix shifts toward a much bigger AI opportunity.

Qualcomm missed the first wave of the AI trade, but it can still be an AI winner...

Qualcomm's AI business is young... Its data-center products are just beginning to ramp up.

That said, the company already has a major hyperscale customer in Meta... which signed a long-term deal to buy Qualcomm's brand-new chip line.

Management has nearly doubled its revenue expectations for non-smartphone chips for 2029. That tells us the company's growth will outpace investors' expectations.

They've spent years learning to discount Qualcomm.

But now, it's staking a claim in an entirely new market... and a surge in profitability could soon follow.

Regards,

Joel Litman


Editor's note: A signal that preceded the rise of each member of the Magnificent Seven is now triggering for a new group of stocks that could soar 1,000% or more thanks to Elon Musk's biggest venture yet... a major project in rural Texas. Yesterday, Joel went on camera to share all the details. If you missed it, check out the replay here.

Further Reading

Apple helped put Qualcomm on the map by kick-starting the smartphone era. And today, Apple is also expanding its operations... into a specific type of AI that's "always on."

Companies like Meta are pouring buckets of cash into AI. But, historically, tech firms have not been as capital-intensive as they are today. This raises an obvious question: Will the hyperscalers be able to afford all this spending?

Back to Top