Paid for by Qualcomm
As AI agents take on more complex tasks, the devices around you will play a larger role, and the cloud behind them will face growing demands. In the data centers that power the cloud, two challenges stand out. The first is memory bandwidth: moving data quickly enough to keep processors supplied with the information they need to deliver on highly demanding AI workloads. And with rapidly scaling usage, moving and processing data at the lowest power possible has also become increasingly essential.
Qualcomm is addressing both challenges with its new Qualcomm High Bandwidth Compute (HBC) technology. By bonding compute logic and memory together onto 3D-stacked silicon, the technology is designed to deliver what Akash Palkhiwala, Qualcomm’s executive vice president, chief financial officer and chief operating officer, describes as a “very, very high memory bandwidth solution.” Going further, Qualcomm designed HBC to deliver that bandwidth with greater power efficiency than today’s high-bandwidth memory designs, an advantage that could add up across a data center as AI scales.
This unique solution arose from Qualcomm’s position as a leader in mobile computing. Qualcomm has long had to design within tight power, size and heat limits, finding ways to deliver high performance at low-power. HBC is a recent example of that approach, now applied to the data center. Palkhiwala argues that this experience gives Qualcomm a foundation for the expansion.
From the phone to the data center
Qualcomm’s expansion starts with technology it spent decades developing for smartphones. Wireless connectivity, high-performance, low power computing and AI capabilities are now needed for a range of products, from PCs to cars to industrial equipment. “We just found ourselves in this fortunate place where the set of technologies that we’ve been working on for the last 15 years for smartphones is directly applicable,” notes Palkhiwala.
That same high-performance, low-power heritage underpins the new Qualcomm Dragonfly data center portfolio, which brings together power-efficient CPUs, AI accelerators, custom silicon and connectivity. As people and AI agents divide work between devices and the cloud, Qualcomm aims to offer high-performance, power-efficient computing across both.
Computing across devices and the cloud
As AI becomes more pervasive, Qualcomm sees a major opportunity in distributed systems that span devices and the cloud, linked by advanced connectivity. For example, as chatbots evolve into agents that can work through tasks, those agents will need access to computing and information in different places. “The device sometimes has information that the cloud does not,” Palkhiwala says. A phone, for example, can provide context about where someone is or what its camera sees.
Palkhiwala also sees further opportunities in robotics, automotive, industrial AI and new personal AI devices. He envisions “a third device beyond the PC and phone that’s even more personal than the phone.” For Qualcomm, that opens up more places to apply technology it has already developed. “One of the advantages we’ve always had is the technology that we are creating applies to a bunch of end industries,” Palkhiwala says.
From Qualcomm:
Qualcomm is a global computing leader at the center of the AI era, enabling intelligence to scale from the most personal devices to largescale infrastructure. Building on more than four decades of innovation, we develop platforms and solutions that bring together advanced AI, high-performance, low-power computing, and industry leading connectivity—powering products and services used around the world. At Qualcomm, we are engineering human progress. Learn more at qualcomm.com.
Qualcomm and Qualcomm Dragonfly branded products are products of Qualcomm Technologies, Inc. and/or its subsidiaries.
This article was paid for by Qualcomm and created by Yahoo Brand Studio. The Yahoo Finance editorial staff did not participate in the creation of this content.