Qualcomm targets $15B data center chip sales by 2029, shares jump 13%

Qualcomm's push into AI data center chips challenges Nvidia's dominance, backed by design wins at Microsoft and Meta.

Qualcomm expects to generate $15 billion in data center chip revenue by fiscal 2029, a push into AI infrastructure that threatens Nvidia's hold on the market for artificial intelligence processors.

"We will be truly diversified," Chief Financial Officer Akash Palkhiwala said at an investor presentation Wednesday, projecting data center chips will bring in $5 billion by fiscal 2027.

Of that $5 billion target, $1 billion will come from two unnamed hyperscaler customers for whom Qualcomm is building custom chips, with revenue starting before year-end. Microsoft will use Qualcomm's new High Bandwidth Compute chips for AI inference tasks, while Meta will deploy the Dragonfly C1000 CPU designed for AI data centers. The company also raised its long-term revenue target for chips outside its smartphone business to $40 billion by 2029, nearly double its prior estimate of $22 billion.

Qualcomm shares surged more than 12% in after-hours trading, while Arm Holdings, which supplies underlying technology for many Qualcomm chips, rose 5%. The push comes as Qualcomm's handset revenue fell 13% year over year in the prior quarter, squeezed by a memory chip shortage and major customers such as Apple and Samsung developing chips in-house.

A New Chip Category Targets Cost Advantage

Qualcomm's data center chief Tony Pialis described the High Bandwidth Compute chips as a new category that uses cheaper memory found in smartphones and laptops, rather than the expensive high-bandwidth memory used by Nvidia or the SRAM memory employed by Cerebras Systems. "That is a tremendous value that we deliver to the industry in terms of performance per cost advantage," Pialis said.

The Dragonfly C1000 CPU, which Meta will deploy, enters a market where both Arm Holdings and Nvidia are courting data center customers. Qualcomm said it is working with customers on three types of chips: central processing units, inference accelerators, and custom application-specific integrated circuits, a segment where Broadcom and Marvell have seen booming demand.

Crowded Field, High Stakes

Qualcomm is re-entering a fast-growing but hyper-competitive AI chip market dominated by Nvidia, the newly public Cerebras, and custom chip options from Amazon's Graviton and Google's Axion. Bank of America analysts had projected modest revenue of roughly $2 billion to $5 billion annually from Qualcomm's data center push by fiscal 2027-2028, placing the company's $5 billion target at the high end of expectations.

Pialis said he has not had to push his way into hyperscale customers. "They've been pulling us in," he said, without naming the two custom-chip clients.

Qualcomm shares gained about 13% in pre-market trading following the announcement. Morgan Stanley upgraded the stock to Equalweight with a $231 price target. If Qualcomm delivers on its $15 billion data center revenue target by 2029, it would represent roughly 15% of Nvidia's current annual data center revenue, showing that while the gap remains wide, Qualcomm is becoming a credible alternative for hyperscalers seeking to diversify chip suppliers.

This article is for informational purposes only and does not constitute investment advice.