Inquirer EntertainmentJC Santos plays abducted priest in ‘Romano,’ highlighting plight of missing activistsThe Jerusalem PostWhat is the ICJ, the UN body that just ruled Myanmar committed genocide? - explainerInquirerLTO suspends lady rider’s license over ‘dangerous stunt’ in CaviteDaily MaverickCommunity Innovation: From polluted canals to new horizons — imagining Gqeberha’s futureZDF heuteEntdecken Sie das ZDF-NachrichtenstudioPremium TimesFormer NUJ Abuja chairman, Abdul Jelil Adebayo, dies at 62VanguardTinubu: First 100 Yutong CNG buses to arrive DecemberCBS NewsAnthropic set to start trading after completing largest-ever IPORai NewsStorico verdetto della Corte internazionale di giustizia: "In Myanmar contro Rohingya fu genocidio"NumeramaUn mois après le chaos PHP2Shell, que vaut vraiment la nouvelle alerte critique de PHP ?RMF24Najdłuższy tunel drogowy w Polsce. To oni zbudują maszynę, która go wydrążyDW DeutschUN-Gericht: Myanmar verantwortlich für Rohingya-Völkermord
The Daily Newsstand · Free, Always
Thursday, November 19, 2026

Qualcomm offers test drive of its rack-scale Nvidia AI challenger

Translate

Selected customers can put Dragonfly AI200 through its paces on Cirrascale's cloud

Qualcomm has opened cloud-based trials of its Dragonfly AI200 rack-scale inference system to selected customers through Cirrascale Cloud Services.

Dragonfly AI200 is Qualcomm's pitch for running large language, multimodal, and other AI models without Nvidia-sized power and operating bills.

Whether it delivers those savings is something customers can now put to the test on Cirrascale's AI Innovation Cloud, with a route into production deployments, Qualcomm says. Cirrascale also demonstrated the system at SC26 in Chicago on November 17-18.

REG AD

The AI200 is part of Qualcomm's Dragonfly datacenter lineup, which also includes the AI250 and AI300 accelerators and the C1000 server CPU. The company set out its broader ambitions at its Investor Day in June.

REG AD

Each AI200 accelerator card carries 768 GB of LPDDR memory, with 56 cards providing about 43 TB in a liquid-cooled rack. Qualcomm says the system supports models ranging from 7 billion to 10 trillion parameters, including long-context, retrieval-augmented generation (RAG), and agentic workloads.

As we previously explored, packing in memory doesn't necessarily mean matching Nvidia's performance. Its Blackwell-based NVL72 systems use fast – and costly – HBM3e stacks, while Qualcomm has opted for LPDDR, more commonly found in mobile devices.

"AI inference is becoming one of the largest and most important computing workloads in the datacenter, and the industry needs a more efficient approach," said Tony Pialis, Qualcomm's executive vice president and general manager for datacenter.

He claimed that the AI200 brings the company's experience with high-performance, low-power computing to a rack-scale infrastructure, in an architecture designed around the performance, memory, and cost requirements of inference workloads.

"Our customers need efficient, scalable infrastructure that can support increasingly sophisticated models without allowing power and operating costs to become the limiting factor," said Cirrascale CEO Dave Driggers.

"By working with Qualcomm, we can give customers evaluation access to a compelling new architecture purpose-built for the next generation of AI inference."

The chip firm also has its own software stack, the Dragonfly AI Infrastructure Management Suite, for provisioning, monitoring, workload orchestration, and fault handling across rack-scale deployments. ®

View the original on The Register →

KioskNews shows a cleaned-up reading view extracted from the publisher’s page — the original always lives on their site, not ours.