Is Qualcomm the New AI Chip Growth Stock to Buy?

"Qualcomm ( NASDAQ:QCOM ) has long dominated the mobile chip market, but its latest push into artificial intelligence (AI) hardware raises questions for investors. With the AI boom driving massive gains for leaders like Nvidia ( NASDAQ:NVDA ) and Advanced Micro Devices ( ), Qualcomm's announcement of two new data center chips aims to carve out a niche. These chips target AI inference tasks, where trained models process data in real time."

"Earlier this week, Qualcomm unveiled its new AI200 and AI250 chips, marking a significant expansion beyond smartphones. The AI200, slated for commercial availability in 2026, focuses on enhanced memory capacity and efficient AI inference. It supports up to 768 gigabytes of memory per rack, surpassing current offerings from competitors, and is designed for liquid-cooled server systems that can scale to 72 chips acting as one unit."

"The AI250 follows in 2027, building on this with even more advanced memory architecture for near-memory computing, promising better performance in handling large AI models. These chips leverage Qualcomm's Hexagon neural processing units, adapted from mobile tech, to emphasize power efficiency. A full rack consumes about 160 kilowatts, comparable to some Nvidia setups but with claims of lower operating costs."

Qualcomm unveiled AI200 and AI250 data-center chips targeting AI inference workloads. The AI200 will be commercially available in 2026, supports up to 768 gigabytes of memory per rack, and is designed for liquid-cooled servers that can scale to 72 chips acting as one unit. The AI250, expected in 2027, adds advanced memory architecture for near-memory computing to improve performance on large models. The chips use Hexagon neural processing units adapted from mobile designs to emphasize power efficiency. A full rack consumes about 160 kilowatts, and Qualcomm offers chips or complete racks supporting TensorFlow and PyTorch. Qualcomm also signed a deal with Saudi Arabia's Humain to deploy 200 megawatts of AI infrastructure starting in 2026.

Read at 24/7 Wall St.