At Qualcomm's investor day presentation, it was looking to accelerate CEO Cristiano Amon's long term strategy to diversify the chip maker away from its core market, smartphones.
This week's announcements will provide meaningful revenue as soon as next fiscal year, and bring the company one step closer to making smartphone chips only a third of its sales by 2029.
Jump-started by Qualcomm's 2021 Nuvia acquisition, the company's highly regarded chips now run numerous consumer goods besides smartphones, like the expansive Mercedes-Benz CLA infotainment screens, Meta Platforms smartglasses, and Dell laptops.
Non-smartphone chips made up 28% of total chip revenue last fiscal year, which ended in September. Qualcomm has slowly become a more diversified chip vendor.
Now it's trying to speed up the project yet again with acquisitions. This time the target isn't consumer devices, but hundreds of billions of dollars in artificial-intelligence data center capital expenditures.
Among chip makers, Nvidia was the big early AI winner and remains so. It was ready for the AI explosion, having already spent 15 years building an ecosystem made of four key parts: AI accelerator chips, CPUs, high-speed networking chips, and mature software that tied it all together. This complete "stack" of technologies is why it's been so difficult for competitors to dislodge Nvidia from its pole position.
But things have changed. Creating new bigger and better AI models was the predominant use for AI computing, but now running AI models, a process known as "inference," is starting to become the bigger workload.
That's set to accelerate in the coming years as enterprises begin to deploy more agents. Agents are software that can use AI models to accomplish a complex series of tasks from simple conversational commands. An April study from researchers at Google, Microsoft, and top universities found that for its main current use, software coding, agents use a thousand times more inference than humans.
As inference takes over, Nvidia's competitors see an opportunity to edge into this giant and fast growing market. Moreover, its customers have a chance to diversify their suppliers, including by making their own custom chips.
Now Qualcomm is jumping into this crowded pool, and in a relatively short period of time, the chip maker has put together an AI data center tech stack that looks a little like Nvidia's. We will see these products begin to roll out in Qualcomm's fiscal 2027, building over the following two years.
When it is done ramping, Qualcomm will have all four pieces of the puzzle. First, it has a high speed inference chip, the HBC platform. Microsoft may be a customer, but details were thin.
"HBC implements an innovative architecture with high memory bandwidth and integrated compute that unlocks significant improvements in cost and performance for the next generation of AI infrastructure," said Microsoft CEO Satya Nadella in a video during Qualcomm's investor day presentation on Wednesday.
The first HBC generation will be available for customer sampling next year, and the second generation will follow in 2028.
That same year, Qualcomm will also release a CPU--a strength for Qualcomm since the Nuvia acquisition--with Meta as the first data center customer.
In December, Qualcomm acquired Alphawave Semi, which brought in high-speed data center connectivity chips and a custom silicon design lab. One of the startup's founders, Tony Pialis, is now executive vice president of data center tech at Qualcomm. The networking chips are already for sale, and Qualcomm has two custom-chip customers signed on. These will be contributing to Qualcomm sales next fiscal year.
Finally, on Wednesday Qualcomm announced the acquisition of Modular, which brings software for AI inference and programming. The deal also includes the startup's co-founder and CEO, Chris Lattner, a legendary figure in software circles.
This puts all the key pieces together. Qualcomm is a little late to the table, and it will take a few years for the meal to be served, but it looks like a feast.
Write to Adam Levine at adam.levine@barrons.com
This content was created by Barron's, which is operated by Dow Jones & Co. Barron's is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
June 28, 2026 03:00 ET (07:00 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments