Qualcomm has confirmed its long-rumored data center CPU rollout will start in collaboration with Saudi Arabia's state-backed AI cloud infrastructure project, in partnership with HUMAIN AI. In a press release, the company confirmed it has signed a memorandum of understanding with HUMAIN, a new Saudi Arabian venture backed by the country's Public Investment Fund (PIF).
The move represents a return to making data center CPUs after Qualcomm's first failed initiative. Qualcomm famously developed Centriq data center CPUs in 2017 but then canceled the project in 2019. The company later purchased the startup Nuvia, which was designing Arm-based data center CPUs. Qualcomm repurposed the tech for its Snapdragon processors for Windows machines, but speculation has continued that the company would re-enter the data center CPU market with Nuvia's tech under the hood.
Qualcomm and HUMAIN AI say they will partner in developing next-generation AI data centers, infrastructure, and cloud-to-edge services. Specifically, Qualcomm says it intends to develop and build "cutting-edge AI data centers in Saudi Arabia designed to deliver highly efficient, scalable cloud-to-edge hybrid AI inferencing solutions for local and international customers based on Qualcomm Technologies’ edge and data center solutions."
Qualcomm says it will also develop and supply state-of-the-art data center CPU and AI solutions to HUMAIN's data centers and speed up the use of the infrastructure by leveraging Snapdragon and Dragonwing processors.
Qualcomm and HUMAIN say their data centers "are intended to provide both government and enterprise organizations with access to high-performance and power efficient CPU and AI cloud infrastructure and cloud-to-edge services."
As part of the agreement, Qualcomm will also establish a Qualcomm Design Center in Saudi Arabia for semiconductor technology.
Cristiano Amon, President and CEO of Qualcomm Incorporated, said Qualcomm was "pleased to be working with HUMAIN to support the realization" of its goal to become a global AI leader.
Qualcomm hinted at plans to develop server CPUs earlier this year when it hired Sailesh Kottapalli, former chief architect of Intel's Xeon server processors. Kottapalli's tenure included projects such as x86, Itanium, and multiple Xeon generations.
The news follows Nvidia's own announcement that it will send 18,000 AI GPUs to the country, which will be used to build a 500-megawatt data center. It follows the Trump administration's official scrapping of the AI Diffusion Rule, which restricted the export of AI chips to most countries.
Follow Tom's Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.
