Qualcomm Unveils 3-Part AI Server Push With 80-Core Oryon Chips

1 articles · Updated · Barchart · May 24

Qualcomm outlined a three-part server strategy built around 80-core Oryon CPUs, AI200/AI250 inference accelerators and custom cloud chips, marking a direct push into AI infrastructure.
Inference workloads are becoming the main task for cloud providers, and Qualcomm is betting its low-power chip design can cut data-center energy use and total cost of ownership better than hotter GPU-heavy systems.
An unnamed large customer is already working with Qualcomm on AI chip deliveries due by the end of 2026, signaling demand for alternatives to expensive, power-hungry incumbents.
Beyond servers, Qualcomm is extending the same efficiency pitch into robotics through its Dragonwing IQ10 platform and a March 2026 integration deal with NEURA Robotics.
The broader bet is that Qualcomm can reduce its reliance on smartphones and Apple by becoming a standard supplier for AI inference and eventually physical AI devices.

Can Qualcomm's energy-efficient AI chips break NVIDIA's stronghold on the lucrative data center market?

With its new robotics platform, how soon will Qualcomm-powered humanoids be working alongside us?