Stock Markets June 26, 2026 01:03 PM

Qualcomm to Adapt Data Center Chip Architecture for Mobile Devices

Company says vertically stacked High Bandwidth Compute design will begin in data centers, with potential to bring enhanced on-device AI to smartphones and other devices

By Nina Shah
Share
Twitter Reddit Facebook LinkedIn
QCOM

Qualcomm announced plans to apply chip technology it developed for data centers to smartphones and other devices, aiming to improve on-device artificial intelligence performance. The company unveiled a vertically stacked High Bandwidth Compute architecture and said the first generation will launch in data centers next year, with commercial availability expected in 2028. Qualcomm is engaged in talks with device makers but has not provided a timetable for bringing the technology to handhelds.

Qualcomm to Adapt Data Center Chip Architecture for Mobile Devices
QCOM
Summarize with
ChatGPT Perplexity Claude Grok Gemini

Key Points

  • Qualcomm plans to reuse data center chip technology in smartphones and other devices to enhance on-device AI.
  • The company's High Bandwidth Compute architecture stacks chips vertically to place memory and compute closer together, improving data speed and flow.
  • First-generation deployment is slated for data centers next year with commercial availability expected in 2028; discussions are underway with smartphone, PC, and automotive manufacturers.

Overview

Qualcomm said it intends to carry over chip technology created for data centers into smartphones and other consumer devices, positioning the design to boost how artificial intelligence runs locally on mobile hardware. The company revealed the plan through comments by Executive Vice President Durga Malladi in an interview with Semafor.

New data center push

The chip designer formally entered the data center chip market this week by announcing a new line of data center processors. Malladi noted the sector is currently contending with supply shortages and robust competition. "What starts in data centers is not going to end there," she said, describing the intention to extend developments beyond the cloud.

Technology details

At the core of Qualcomm's announcement is a High Bandwidth Compute architecture that stacks chips vertically rather than arranging them side by side. This stacked configuration brings memory and compute components into closer physical proximity, which the company says improves data speed and flow within the chip package.

Qualcomm is already in discussions with manufacturers of smartphones, personal computers, and cars about the new data center technology portfolio, according to Malladi. While the firm has emphasized outreach to multiple device categories, Malladi did not specify when the High Bandwidth Compute approach would be integrated into smartphones or other non-data-center products.

Timeline and rollout

The company expects the first generation of the new architecture to launch in data centers next year, and it anticipates commercial availability in 2028. No timing was provided for adoption in handheld devices.

Implications for mobile AI

Chip stacking is currently concentrated in data center environments rather than in smartphones. Qualcomm described a potential path where, if the technology reaches handheld devices, users could run a greater number of AI models locally and operate agents in an "always on" mode without draining the battery. The company had lagged in data center technology for several years but retains a strong position owing to its long history producing smartphone chips.


Quote

"What starts in data centers is not going to end there," Malladi said, signaling Qualcomm's intent to move innovations across device categories.

Risks

  • Supply shortages and strong competition in the data center chip sector could affect production and deployment timelines - impacting semiconductor and data center markets.
  • Qualcomm has not provided a timetable for integrating the new architecture into smartphones and other devices, creating uncertainty for device makers and mobile AI adoption.
  • Chip stacking is mainly used in data centers today, so technical or integration challenges may delay bringing the approach to handheld devices - affecting smartphone and consumer electronics sectors.

More from Stock Markets

Analysts Shift More Bullish on American Tower, Citing Organic Growth and Data Center Upside Jun 26, 2026 Tech Gains Tempered by Defensive Rotation as Memory, Crypto-Linked Stocks Diverge Jun 26, 2026 Seagate, Marvell Headline Friday’s Sharp Market Movers as Biotech and Software Stocks Also Swing Jun 26, 2026 AeroVironment and Concentrix Headline Friday's Earnings Releases Jun 26, 2026 Dallas Fed manufacturing reading, Treasury bill auctions headline a light Monday data slate Jun 26, 2026