HUMAIN, Qualcomm to deploy AI inferencing for hybrid AI

Data centre will provide world’s first fully optimised edge-to-cloud hybrid AI system

#SaudiArabia #datacentres - HUMAIN, the national artificial intelligence company formed by Public Investment Fund (PIF) and US semiconductor and wireless technology company Qualcomm Technologies have announced plans to deploy advanced AI infrastructure for inferencing and hybrid AI. Targeting 200 megawatts (MW) in 2026, Qualcomm AI200 and AI250 rack solutions will deliver high-performance AI inference services in the Kingdom and globally. Announced ahead of the Future Investment Initiative conference, the news was synchronised with the global launch of the AI200 and AI250. The collaboration will establish what the companies describe as the world’s first fully optimised edge-to-cloud hybrid AI system, positioning Saudi Arabia as a global hub for artificial intelligence.

SO WHAT? - In May, during the Saudi-US Investment Forum, HUMAIN and Qualcomm signed a Memorandum of Understanding to develop next-generation AI data centres and cloud-to-edge services. The strategic agreement covered delivery of cutting-edge AI infrastructure and integration of HUMAIN’s Arabic large language models across Qualcomm-powered edge devices. The new hybrid AI data centre will use the new AI200 and AI250 accelerators—also announced this week — Qualcomm’s first full-scale entry into the data center AI race. Today’s announcement confirms that the two companies will develop an integrated cloud-to-edge AI platform and it will use Qualcomm’s new nextgen chipsets.

Here are some key details about this announcement:

• HUMAIN and Qualcomm Technologies have announced plans to deploy advanced AI infrastructure for inferencing and hybrid AI, to provide 200MW during 2026.

• The initiative will deploy Qualcomm AI200 and AI250 rack solutions starting in 2026, designed to deliver high-performance AI inference services with rack-scale performance and superior memory capacity for fast generative AI inference.

• Qualcomm AI250, which is expected to be available from 2027, introduces an innovative memory architecture, offering a generational leap in effective memory bandwidth and efficiency for AI workloads.

• Qualcomm AI200 and AI250 solutions provide industry-leading total cost of ownership for inference workloads, marking what the companies describe as a major advancement in enabling scalable, efficient and flexible generative AI across industries through fully optimised edge-to-cloud services.

• The partnership will integrate HUMAIN’s Saudi-developed AI models, including the ALLaM multimodal large language model described as one of the world’s most advanced Arabic AI systems, with Qualcomm’s AI platforms to deliver customer-specific solutions.

• Qualcomm and HUMAIN plan to develop tailored solutions addressing specific needs of enterprises and government organisations across the Kingdom and beyond, supporting Saudi Arabia’s objective to establish itself as a global leader in AI innovation.

• The collaboration was announced ahead of the ninth edition of the Future Investment Initiative conference, following an initial announcement made by both companies at the US-Saudi Investment Forum in May 2025.

• The initiative combines HUMAIN’s regional infrastructure and full AI stack capabilities including Saudi-developed AI models with Qualcomm Technologies’ semiconductor innovation, enabling enterprises and government organisations to harness AI at scale with industry-leading performance per total cost of ownership.

ZOOM OUT - Last month HUMAIN unveiled its Horizon Pro AI laptop at Qualcomm’s Snapdragon Summit in Hawaii. The reveal marked more than just a product launch, it was a strategic signal that the Agentic AI era has not only begun, but will become integral to personal computers. The device’s ability to integrate Saudi Arabia’s ALLAM Arabic-first large language model for local AI processing, while leveraging hybrid cloud performance, showcased a future in which regional ecosystems create globally competitive AI hardware and software stacks. Today’s HUMAIN-Qualcomm announcement revealed the next generation cloud infrastructure that will enable fully optimised edge-to-cloud, end-to-end hybrid AI services.

Qualcomm’s AI200 & AI250

The San Diego semiconductor company today unveiled its next-generation AI200 and AI250 data centre AI inference platforms, in a move that will see it compete directly with NVIDIA. Qualcomm stock rose 15% on the news of the new accelerators and the confirmation of the HUMAIN data centre deal.

The new chipsets are engineered to deliver rack-scale performance and superior memory capacity for large-scale generative AI workloads. The AI200 offers up to 768 GB of LPDDR memory per card for cost-efficient, high-performance inference, while the will AI250 introduce an innovative near-memory computing architecture, delivering over 10x higher effective memory bandwidth and markedly lower power consumption. Designed for secure, scalable deployment, both systems integrate liquid cooling, confidential computing, and broad compatibility with leading AI frameworks.

[Written and edited with the assistance of AI]

Read more about HUMAIN and Qualcomm:

• Saudi HUMAIN unveils revolutionary Agentic AI PC (Middle East AI News)

• HUMAIN to launch Saudi-designed AI PCs in October (Middle East AI News)

• Qualcomm and HUMAIN forge Saudi AI data centre deal (Middle East AI News)

• Qualcomm & SDAIA launch ALLaM AI PC at LEAP25 (Middle East AI News)