Qualcomm has released its latest AI200 and AI250 datacenter chips designed to provide more efficient, and faster generative AI inference at rack scale.
Solutions, the company explains, provide industry-leading performance per dollar and per watt, revolutionizing the deployment of large language and multimodal models in the enterprise space.
The AI200 features up to 768 GB of memory per card for high-end AI applications, and the AI250 offers a new near-memory computing architecture with more than 10 times greater effective memory bandwidth and significantly reduced power consumption.
Both chipsets are built for secure, scalable AI deployment, including confidential computing, direct liquid cooling, and support for top machine learning frameworks.
One-click model deployment with Hugging Face and other platforms is enabled by Qualcomm’s new AI software stack, simplifying integration for developers.
Planned for commercial availability in 2026 and 2027, the AI200 and AI250 are Qualcomm’s most substantial step yet toward building an annual roadmap of high-performance, energy-efficient AI infrastructure for data centers.
You may also want to check out some of our other recent updates.
Wanna know what’s trending online every day? Subscribe to Vavoza Insider to access the latest business and marketing insights, news, and trends daily with unmatched speed and conciseness! 🗞️
